id stringlengths 40 40 | pid stringlengths 42 42 | input stringlengths 12.1k 56k | output stringlengths 2 232 | document_start_index int64 328 328 | document_end_index int64 12k 55.8k | query_start_index int64 12k 55.8k | query_end_index int64 12.1k 55.9k | truncation_seperator stringclasses 1 value |
|---|---|---|---|---|---|---|---|---|
3fad42be0fb2052bb404b989cc7d58b440cd23a0 | 3fad42be0fb2052bb404b989cc7d58b440cd23a0_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Suppose a user wants to write a sentence “I will be 10 minutes late.” Ideally, she would type just a few keywords such as “10 minutes late” and an autocomplete system would be able to infer the intended sentence (Figure FIGREF1). Existing left-to-right autocomplete systems BIBREF0, BIBREF1 can often be inefficient, as the prefix of a sentence (e.g. “I will be”) fails to capture the core meaning of the sentence. Besides the practical goal of building a better autocomplete system, we are interested in exploring the tradeoffs inherent to such communication schemes between the efficiency of typing keywords, accuracy of reconstruction, and interpretability of keywords.
One approach to learn such schemes is to collect a supervised dataset of keywords-sentence pairs as a training set, but (i) it would be expensive to collect such data from users, and (ii) a static dataset would not capture a real user's natural predilection to adapt to the system BIBREF2. Another approach is to avoid supervision and jointly learn a user-system communication scheme to directly optimize the combination of efficiency and accuracy. However, learning in this way can lead to communication schemes that are uninterpretable to humans BIBREF3, BIBREF4 (see Appendix for additional related work).
In this work, we propose a simple, unsupervised approach to an autocomplete system that is efficient, accurate, and interpretable. For interpretability, we restrict keywords to be subsequences of their source sentences based on the intuition that humans can infer most of the original meaning from a few keywords. We then apply multi-objective optimization approaches to directly control and achieve desirable tradeoffs between efficiency and accuracy.
We observe that naively optimizing a linear combination of efficiency and accuracy terms is unstable and leads to suboptimal schemes. Thus, we propose a new objective which optimizes for communication efficiency under an accuracy constraint. We show this new objective is more stable and efficient than the linear objective at all accuracy levels.
As a proof-of-concept, we build an autocomplete system within this framework which allows a user to write sentences by specifying keywords. We empirically show that our framework produces communication schemes that are 52.16% more accurate than rule-based baselines when specifying 77.37% of sentences, and 11.73% more accurate than a naive, weighted optimization approach when specifying 53.38% of sentences. Finally, we demonstrate that humans can easily adapt to the keyword-based autocomplete system and save nearly 50% of time compared to typing a full sentence in our user study.
Approach
Consider a communication game in which the goal is for a user to communicate a target sequence $x= (x_1, ..., x_m)$ to a system by passing a sequence of keywords $z= (z_1, ..., z_n)$. The user generates keywords $z$ using an encoding strategy $q_{\alpha }(z\mid x)$, and the system attempts to guess the target sequence $x$ via a decoding strategy $p_{\beta }(x\mid z)$.
A good communication scheme $(q_{\alpha }, p_{\beta })$ should be both efficient and accurate. Specifically, we prefer schemes that use fewer keywords (cost), and the target sentence $x$ to be reconstructed with high probability (loss) where
Based on our assumption that humans have an intuitive sense of retaining important keywords, we restrict the set of schemes to be a (potentially noncontiguous) subsequence of the target sentence. Our hypothesis is that such subsequence schemes naturally ensure interpretability, as efficient human and machine communication schemes are both likely to involve keeping important content words.
Approach ::: Modeling with autoencoders.
To learn communication schemes without supervision, we model the cooperative communication between a user and system through an encoder-decoder framework. Concretely, we model the user's encoding strategy $q_{\alpha }(z\mid x)$ with an encoder which encodes the target sentence $x$ into the keywords $z$ by keeping a subset of the tokens. This stochastic encoder $q_{\alpha }(z\mid x)$ is defined by a model which returns the probability of each token retained in the final subsequence $z$. Then, we sample from Bernoulli distributions according to these probabilities to either keep or drop the tokens independently (see Appendix for an example).
We model the autocomplete system's decoding strategy $p_{\beta }(x\mid z)$ as a probabilistic model which conditions on the keywords $z$ and returns a distribution over predictions $x$. We use a standard sequence-to-sequence model with attention and copying for the decoder, but any model architecture can be used (see Appendix for details).
Approach ::: Multi-objective optimization.
Our goal now is to learn encoder-decoder pairs which optimally balance the communication cost and reconstruction loss. The simplest approach to balancing efficiency and accuracy is to weight $\mathrm {cost}(x, \alpha )$ and $\mathrm {loss}(x, \alpha , \beta )$ linearly using a weight $\lambda $ as follows,
where the expectation is taken over the population distribution of source sentences $x$, which is omitted to simplify notation. However, we observe that naively weighting and searching over $\lambda $ is suboptimal and highly unstable—even slight changes to the weighting results in degenerate schemes which keep all or none of its tokens. This instability motivates us to develop a new stable objective.
Our main technical contribution is to draw inspiration from the multi-objective optimization literature and view the tradeoff as a sequence of constrained optimization problems, where we minimize the expected cost subject to varying expected reconstruction error constraints $\epsilon $,
This greatly improves the stability of the training procedure. We empirically observe that the model initially keeps most of the tokens to meet the constraints, and slowly learns to drop uninformative words from the keywords to minimize the cost. Furthermore, $\epsilon $ in Eq (DISPLAY_FORM6) allows us to directly control the maximum reconstruction error of resulting schemes, whereas $\lambda $ in Eq (DISPLAY_FORM5) is not directly related to any of our desiderata.
To optimize the constrained objective, we consider the Lagrangian of Eq (DISPLAY_FORM6),
Much like the objective in Eq (DISPLAY_FORM5) we can compute unbiased gradients by replacing the expectations with their averages over random minibatches. Although gradient descent guarantees convergence on Eq (DISPLAY_FORM7) only when the objective is convex, we find that not only is the optimization stable, the resulting solution achieves better performance than the weighting approach in Eq (DISPLAY_FORM5).
Approach ::: Optimization.
Optimization with respect to $q_{\alpha }(z\mid x)$ is challenging as $z$ is discrete, and thus, we cannot differentiate $\alpha $ through $z$ via the chain rule. Because of this, we use the stochastic REINFORCE estimate BIBREF5 as follows:
We perform joint updates on $(\alpha , \beta , \lambda )$, where $\beta $ and $\lambda $ are updated via standard gradient computations, while $\alpha $ uses an unbiased, stochastic gradient estimate where we approximate the expectation in Eq (DISPLAY_FORM9). We use a single sample from $q_{\alpha }(z\mid x)$ and moving-average of rewards as a baseline to reduce variance.
Experiments
We evaluate our approach by training an autocomplete system on 500K randomly sampled sentences from Yelp reviews BIBREF6 (see Appendix for details). We quantify the efficiency of a communication scheme $(q_{\alpha },p_{\beta })$ by the retention rate of tokens, which is measured as the fraction of tokens that are kept in the keywords. The accuracy of a scheme is measured as the fraction of sentences generated by greedily decoding the model that exactly matches the target sentence.
Experiments ::: Effectiveness of constrained objective.
We first show that the linear objective in Eq (DISPLAY_FORM5) is suboptimal compared to the constrained objective in Eq (DISPLAY_FORM6). Figure FIGREF10 compares the achievable accuracy and efficiency tradeoffs for the two objectives, which shows that the constrained objective results in more efficient schemes than the linear objective at every accuracy level (e.g. 11.73% more accurate at a 53.38% retention rate).
We also observe that the linear objective is highly unstable as a function of the tradeoff parameter $\lambda $ and requires careful tuning. Even slight changes to $\lambda $ results in degenerate schemes that keep all or none of the tokens (e.g. $\lambda \le 4.2$ and $\lambda \ge 4.4$). On the other hand, the constrained objective is substantially more stable as a function of $\epsilon $ (e.g. points for $\epsilon $ are more evenly spaced than $\lambda $).
Experiments ::: Efficiency-accuracy tradeoff.
We quantify the efficiency-accuracy tradeoff compared to two rule-based baselines: Unif and Stopword. The Unif encoder randomly keeps tokens to generate keywords with the probability $\delta $. The Stopword encoder keeps all tokens but drops stop words (e.g. `the', `a', `or') all the time ($\delta =0$) or half of the time ($\delta =0.5$). The corresponding decoders for these encoders are optimized using gradient descent to minimize the reconstruction error (i.e. $\mathrm {loss}(x, \alpha , \beta )$).
Figure FIGREF10 shows that two baselines achieve similar tradeoff curves, while the constrained model achieves a substantial 52.16% improvement in accuracy at a 77.37% retention rate compared to Unif, thereby showing the benefits of jointly training the encoder and decoder.
Experiments ::: Robustness and analysis.
We provide additional experimental results on the robustness of learned communication schemes as well as in-depth analysis on the correlation between the retention rates of tokens and their properties, which we defer to Appendix and for space.
Experiments ::: User study.
We recruited 100 crowdworkers on Amazon Mechanical Turk (AMT) and measured completion times and accuracies for typing randomly sampled sentences from the Yelp corpus. Each user was shown alternating autocomplete and writing tasks across 50 sentences (see Appendix for user interface). For the autocomplete task, we gave users a target sentence and asked them to type a set of keywords into the system. The users were shown the top three suggestions from the autocomplete system, and were asked to mark whether each of these three suggestions was semantically equivalent to the target sentence. For the writing task, we gave users a target sentence and asked them to either type the sentence verbatim or a sentence that preserves the meaning of the target sentence.
Table TABREF13 shows two examples of the autocomplete task and actual user-provided keywords. Each column contains a set of keywords and its corresponding top three suggestions generated by the autocomplete system with beam search. We observe that the system is likely to propose generic sentences for under-specified keywords (left column) and almost the same sentences for over-specified keywords (right column). For properly specified keywords (middle column), the system completes sentences accordingly by adding a verb, adverb, adjective, preposition, capitalization, and punctuation.
Overall, the autocomplete system achieved high accuracy in reconstructing the keywords. Users marked the top suggestion from the autocomplete system to be semantically equivalent to the target $80.6$% of the time, and one of the top 3 was semantically equivalent $90.11$% of the time. The model also achieved a high exact match accuracy of 18.39%. Furthermore, the system was efficient, as users spent $3.86$ seconds typing keywords compared to $5.76$ seconds for full sentences on average. The variance of the typing time was $0.08$ second for keywords and $0.12$ second for full sentences, indicating that choosing and typing keywords for the system did not incur much overhead.
Experiments ::: Acknowledgments
We thank the reviewers and Yunseok Jang for their insightful comments. This work was supported by NSF CAREER Award IIS-1552635 and an Intuit Research Award.
Experiments ::: Reproducibility
All code, data and experiments are available on CodaLab at https://bit.ly/353fbyn.
Question:
What are the baselines used?
Answer:
| Unif and Stopword | 328 | 12,627 | 12,629 | 12,667 | ... [The rest of the article is omitted]
|
3fad42be0fb2052bb404b989cc7d58b440cd23a0 | 3fad42be0fb2052bb404b989cc7d58b440cd23a0_1 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Suppose a user wants to write a sentence “I will be 10 minutes late.” Ideally, she would type just a few keywords such as “10 minutes late” and an autocomplete system would be able to infer the intended sentence (Figure FIGREF1). Existing left-to-right autocomplete systems BIBREF0, BIBREF1 can often be inefficient, as the prefix of a sentence (e.g. “I will be”) fails to capture the core meaning of the sentence. Besides the practical goal of building a better autocomplete system, we are interested in exploring the tradeoffs inherent to such communication schemes between the efficiency of typing keywords, accuracy of reconstruction, and interpretability of keywords.
One approach to learn such schemes is to collect a supervised dataset of keywords-sentence pairs as a training set, but (i) it would be expensive to collect such data from users, and (ii) a static dataset would not capture a real user's natural predilection to adapt to the system BIBREF2. Another approach is to avoid supervision and jointly learn a user-system communication scheme to directly optimize the combination of efficiency and accuracy. However, learning in this way can lead to communication schemes that are uninterpretable to humans BIBREF3, BIBREF4 (see Appendix for additional related work).
In this work, we propose a simple, unsupervised approach to an autocomplete system that is efficient, accurate, and interpretable. For interpretability, we restrict keywords to be subsequences of their source sentences based on the intuition that humans can infer most of the original meaning from a few keywords. We then apply multi-objective optimization approaches to directly control and achieve desirable tradeoffs between efficiency and accuracy.
We observe that naively optimizing a linear combination of efficiency and accuracy terms is unstable and leads to suboptimal schemes. Thus, we propose a new objective which optimizes for communication efficiency under an accuracy constraint. We show this new objective is more stable and efficient than the linear objective at all accuracy levels.
As a proof-of-concept, we build an autocomplete system within this framework which allows a user to write sentences by specifying keywords. We empirically show that our framework produces communication schemes that are 52.16% more accurate than rule-based baselines when specifying 77.37% of sentences, and 11.73% more accurate than a naive, weighted optimization approach when specifying 53.38% of sentences. Finally, we demonstrate that humans can easily adapt to the keyword-based autocomplete system and save nearly 50% of time compared to typing a full sentence in our user study.
Approach
Consider a communication game in which the goal is for a user to communicate a target sequence $x= (x_1, ..., x_m)$ to a system by passing a sequence of keywords $z= (z_1, ..., z_n)$. The user generates keywords $z$ using an encoding strategy $q_{\alpha }(z\mid x)$, and the system attempts to guess the target sequence $x$ via a decoding strategy $p_{\beta }(x\mid z)$.
A good communication scheme $(q_{\alpha }, p_{\beta })$ should be both efficient and accurate. Specifically, we prefer schemes that use fewer keywords (cost), and the target sentence $x$ to be reconstructed with high probability (loss) where
Based on our assumption that humans have an intuitive sense of retaining important keywords, we restrict the set of schemes to be a (potentially noncontiguous) subsequence of the target sentence. Our hypothesis is that such subsequence schemes naturally ensure interpretability, as efficient human and machine communication schemes are both likely to involve keeping important content words.
Approach ::: Modeling with autoencoders.
To learn communication schemes without supervision, we model the cooperative communication between a user and system through an encoder-decoder framework. Concretely, we model the user's encoding strategy $q_{\alpha }(z\mid x)$ with an encoder which encodes the target sentence $x$ into the keywords $z$ by keeping a subset of the tokens. This stochastic encoder $q_{\alpha }(z\mid x)$ is defined by a model which returns the probability of each token retained in the final subsequence $z$. Then, we sample from Bernoulli distributions according to these probabilities to either keep or drop the tokens independently (see Appendix for an example).
We model the autocomplete system's decoding strategy $p_{\beta }(x\mid z)$ as a probabilistic model which conditions on the keywords $z$ and returns a distribution over predictions $x$. We use a standard sequence-to-sequence model with attention and copying for the decoder, but any model architecture can be used (see Appendix for details).
Approach ::: Multi-objective optimization.
Our goal now is to learn encoder-decoder pairs which optimally balance the communication cost and reconstruction loss. The simplest approach to balancing efficiency and accuracy is to weight $\mathrm {cost}(x, \alpha )$ and $\mathrm {loss}(x, \alpha , \beta )$ linearly using a weight $\lambda $ as follows,
where the expectation is taken over the population distribution of source sentences $x$, which is omitted to simplify notation. However, we observe that naively weighting and searching over $\lambda $ is suboptimal and highly unstable—even slight changes to the weighting results in degenerate schemes which keep all or none of its tokens. This instability motivates us to develop a new stable objective.
Our main technical contribution is to draw inspiration from the multi-objective optimization literature and view the tradeoff as a sequence of constrained optimization problems, where we minimize the expected cost subject to varying expected reconstruction error constraints $\epsilon $,
This greatly improves the stability of the training procedure. We empirically observe that the model initially keeps most of the tokens to meet the constraints, and slowly learns to drop uninformative words from the keywords to minimize the cost. Furthermore, $\epsilon $ in Eq (DISPLAY_FORM6) allows us to directly control the maximum reconstruction error of resulting schemes, whereas $\lambda $ in Eq (DISPLAY_FORM5) is not directly related to any of our desiderata.
To optimize the constrained objective, we consider the Lagrangian of Eq (DISPLAY_FORM6),
Much like the objective in Eq (DISPLAY_FORM5) we can compute unbiased gradients by replacing the expectations with their averages over random minibatches. Although gradient descent guarantees convergence on Eq (DISPLAY_FORM7) only when the objective is convex, we find that not only is the optimization stable, the resulting solution achieves better performance than the weighting approach in Eq (DISPLAY_FORM5).
Approach ::: Optimization.
Optimization with respect to $q_{\alpha }(z\mid x)$ is challenging as $z$ is discrete, and thus, we cannot differentiate $\alpha $ through $z$ via the chain rule. Because of this, we use the stochastic REINFORCE estimate BIBREF5 as follows:
We perform joint updates on $(\alpha , \beta , \lambda )$, where $\beta $ and $\lambda $ are updated via standard gradient computations, while $\alpha $ uses an unbiased, stochastic gradient estimate where we approximate the expectation in Eq (DISPLAY_FORM9). We use a single sample from $q_{\alpha }(z\mid x)$ and moving-average of rewards as a baseline to reduce variance.
Experiments
We evaluate our approach by training an autocomplete system on 500K randomly sampled sentences from Yelp reviews BIBREF6 (see Appendix for details). We quantify the efficiency of a communication scheme $(q_{\alpha },p_{\beta })$ by the retention rate of tokens, which is measured as the fraction of tokens that are kept in the keywords. The accuracy of a scheme is measured as the fraction of sentences generated by greedily decoding the model that exactly matches the target sentence.
Experiments ::: Effectiveness of constrained objective.
We first show that the linear objective in Eq (DISPLAY_FORM5) is suboptimal compared to the constrained objective in Eq (DISPLAY_FORM6). Figure FIGREF10 compares the achievable accuracy and efficiency tradeoffs for the two objectives, which shows that the constrained objective results in more efficient schemes than the linear objective at every accuracy level (e.g. 11.73% more accurate at a 53.38% retention rate).
We also observe that the linear objective is highly unstable as a function of the tradeoff parameter $\lambda $ and requires careful tuning. Even slight changes to $\lambda $ results in degenerate schemes that keep all or none of the tokens (e.g. $\lambda \le 4.2$ and $\lambda \ge 4.4$). On the other hand, the constrained objective is substantially more stable as a function of $\epsilon $ (e.g. points for $\epsilon $ are more evenly spaced than $\lambda $).
Experiments ::: Efficiency-accuracy tradeoff.
We quantify the efficiency-accuracy tradeoff compared to two rule-based baselines: Unif and Stopword. The Unif encoder randomly keeps tokens to generate keywords with the probability $\delta $. The Stopword encoder keeps all tokens but drops stop words (e.g. `the', `a', `or') all the time ($\delta =0$) or half of the time ($\delta =0.5$). The corresponding decoders for these encoders are optimized using gradient descent to minimize the reconstruction error (i.e. $\mathrm {loss}(x, \alpha , \beta )$).
Figure FIGREF10 shows that two baselines achieve similar tradeoff curves, while the constrained model achieves a substantial 52.16% improvement in accuracy at a 77.37% retention rate compared to Unif, thereby showing the benefits of jointly training the encoder and decoder.
Experiments ::: Robustness and analysis.
We provide additional experimental results on the robustness of learned communication schemes as well as in-depth analysis on the correlation between the retention rates of tokens and their properties, which we defer to Appendix and for space.
Experiments ::: User study.
We recruited 100 crowdworkers on Amazon Mechanical Turk (AMT) and measured completion times and accuracies for typing randomly sampled sentences from the Yelp corpus. Each user was shown alternating autocomplete and writing tasks across 50 sentences (see Appendix for user interface). For the autocomplete task, we gave users a target sentence and asked them to type a set of keywords into the system. The users were shown the top three suggestions from the autocomplete system, and were asked to mark whether each of these three suggestions was semantically equivalent to the target sentence. For the writing task, we gave users a target sentence and asked them to either type the sentence verbatim or a sentence that preserves the meaning of the target sentence.
Table TABREF13 shows two examples of the autocomplete task and actual user-provided keywords. Each column contains a set of keywords and its corresponding top three suggestions generated by the autocomplete system with beam search. We observe that the system is likely to propose generic sentences for under-specified keywords (left column) and almost the same sentences for over-specified keywords (right column). For properly specified keywords (middle column), the system completes sentences accordingly by adding a verb, adverb, adjective, preposition, capitalization, and punctuation.
Overall, the autocomplete system achieved high accuracy in reconstructing the keywords. Users marked the top suggestion from the autocomplete system to be semantically equivalent to the target $80.6$% of the time, and one of the top 3 was semantically equivalent $90.11$% of the time. The model also achieved a high exact match accuracy of 18.39%. Furthermore, the system was efficient, as users spent $3.86$ seconds typing keywords compared to $5.76$ seconds for full sentences on average. The variance of the typing time was $0.08$ second for keywords and $0.12$ second for full sentences, indicating that choosing and typing keywords for the system did not incur much overhead.
Experiments ::: Acknowledgments
We thank the reviewers and Yunseok Jang for their insightful comments. This work was supported by NSF CAREER Award IIS-1552635 and an Intuit Research Award.
Experiments ::: Reproducibility
All code, data and experiments are available on CodaLab at https://bit.ly/353fbyn.
Question:
What are the baselines used?
Answer:
| Unif and Stopword | 328 | 12,627 | 12,629 | 12,667 | ... [The rest of the article is omitted]
|
8bf7f1f93d0a2816234d36395ab40c481be9a0e0 | 8bf7f1f93d0a2816234d36395ab40c481be9a0e0_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
This work is licensed under a Creative Commons Attribution 4.0 International License. License details: http://creativecommons.org/licenses/by/4.0/.
Sentence pair modeling is a fundamental technique underlying many NLP tasks, including the following:
Traditionally, researchers had to develop different methods specific for each task. Now neural networks can perform all the above tasks with the same architecture by training end to end. Various neural models BIBREF1 , BIBREF0 , BIBREF10 , BIBREF11 , BIBREF12 , BIBREF13 , BIBREF14 , BIBREF15 have declared state-of-the-art results for sentence pair modeling tasks; however, they were carefully designed and evaluated on selected (often one or two) datasets that can demonstrate the superiority of the model. The research questions are as follows: Do they perform well on other tasks and datasets? How much performance gain is due to certain system design choices and hyperparameter optimizations?
To answer these questions and better understand different network designs, we systematically analyze and compare the state-of-the-art neural models across multiple tasks and multiple domains. Namely, we implement five models and their variations on the same PyTorch platform: InferSent model BIBREF16 , Shortcut-stacked Sentence Encoder Model BIBREF17 , Pairwise Word Interaction Model BIBREF1 , Decomposable Attention Model BIBREF10 , and Enhanced Sequential Inference Model BIBREF0 . They are representative of the two most common approaches: sentence encoding models that learn vector representations of individual sentences and then calculate the semantic relationship between sentences based on vector distance and sentence pair interaction models that use some sorts of word alignment mechanisms (e.g., attention) then aggregate inter-sentence interactions. We focus on identifying important network designs and present a series of findings with quantitative measurements and in-depth analyses, including (i) incorporating inter-sentence interactions is critical; (ii) Tree-LSTM does not help as much as previously claimed but surprisingly improves performance on Twitter data; (iii) Enhanced Sequential Inference Model has the most consistent high performance for larger datasets, while Pairwise Word Interaction Model performs better on smaller datasets and Shortcut-Stacked Sentence Encoder Model is the best performaning model on the Quora corpus. We release our implementations as a toolkit to the research community.
General Framework for Sentence Pair Modeling
Various neural networks have been proposed for sentence pair modeling, all of which fall into two types of approaches. The sentence encoding approach encodes each sentence into a fixed-length vector and then computes sentence similarity directly. The model of this type has advantages in the simplicity of the network design and generalization to other NLP tasks. The sentence pair interaction approach takes word alignment and interactions between the sentence pair into account and often show better performance when trained on in-domain data. Here we outline the two types of neural networks under the same general framework:
Representative Models for Sentence Pair Modeling
Table 1 gives a summary of typical models for sentence pair modeling in recent years. In particular, we investigate five models in depth: two are representative of the sentence encoding type of model, and three are representative of the interaction-aggregation type of model. These models have reported state-or-the-art results with varied architecture design (this section) and implementation details (Section "Implementation Details" ).
The Bi-LSTM Max-pooling Network (InferSent)
We choose the simple Bi-LSTM max-pooling network from InferSent BIBREF16 :
$$&\overleftrightarrow{\mathbf {h}}_{i} = BiLSTM(\mathbf {x}_{i}, \overleftrightarrow{\mathbf {h}}_{i-1}) \\ &\mathbf {v}=max(\overleftrightarrow{\mathbf {h}}_{1}, \overleftrightarrow{\mathbf {h}}_{2}, ..., \overleftrightarrow{\mathbf {h}}_{n})$$ (Eq. 17)
where $\overleftrightarrow{\mathbf {h}}_{i}$ represents the concatenation of hidden states in both directons. It has shown better transfer learning capabilities than several other sentence embedding models, including SkipThought BIBREF31 and FastSent BIBREF32 , when trained on the natural language inference datasets.
The Shortcut-Stacked Sentence Encoder Model (SSE)
The Shortcut-Stacked Sentence Encoder model BIBREF17 is a sentence-based embedding model, which enhances multi-layer Bi-LSTM with skip connection to avoid training error accumulation, and calculates each layer as follows:
$$&\overleftrightarrow{\mathbf {h}}_{i}^{k} = BiLSTM(\mathbf {x}_{i}^{k}, \overleftrightarrow{\mathbf {h}}_{i-1}^{k}) \\ &\mathbf {x}_{i}^{1}=\mathbf {w}_{i} \quad (k=1), \qquad \mathbf {x}_{i}^{k}=[\mathbf {w}_{i}, \overleftrightarrow{\mathbf {h}}_{i}^{k-1}, \overleftrightarrow{\mathbf {h}}_{i}^{k-2}, ..., \overleftrightarrow{\mathbf {h}}_{i}^{1}] \quad (k>1) \\ &\mathbf {v}=max(\overleftrightarrow{\mathbf {h}}_{1}^{m}, \overleftrightarrow{\mathbf {h}}_{2}^{m}, ..., \overleftrightarrow{\mathbf {h}}_{n}^{m})$$ (Eq. 19)
where $\mathbf {x}_{i}^{k}$ is the input of the $k$ th Bi-LSTM layer at time step $i$ , which is the combination of outputs from all previous layers, $\overleftrightarrow{\mathbf {h}}_{i}^{k}$ represents the hidden state of the $k$ th Bi-LSTM layer in both directions. The final sentence embedding $\mathbf {v}$ is the row-based max pooling over the output of the last Bi-LSTM layer, where $n$ denotes the number of words within a sentence and $m$ is the number of Bi-LSTM layers ( $m = 3$ in SSE).
The Pairwise Word Interaction Model (PWIM)
In the Pairwise Word Interaction model BIBREF1 , each word vector $\mathbf {w}_{i}$ is encoded with context through forward and backward LSTMs: $\overrightarrow{\mathbf {h}}_{i} = LSTM^{f}(\mathbf {w}_{i}, \overrightarrow{\mathbf {h}}_{i-1})$ and $\overleftarrow{\mathbf {h}}_{i} = LSTM^{b}(\mathbf {w}_{i}, \overleftarrow{\mathbf {h}}_{i+1})$ . For every word pair $(\mathbf {w}^a_{i}, \mathbf {w}^b_{j})$ across sentences, the model directly calculates word pair interactions using cosine similarity, Euclidean distance, and dot product over the outputs of the encoding layer:
$$D(\overrightarrow{\mathbf {h}}_{i}, \overrightarrow{\mathbf {h}}_{j}) & = [cos(\overrightarrow{\mathbf {h}}_{i}, \overrightarrow{\mathbf {h}}_{j}), \Vert \overrightarrow{\mathbf {h}}_{i} - \overrightarrow{\mathbf {h}}_{j}\Vert , \overrightarrow{\mathbf {h}}_{i} \cdot \overrightarrow{\mathbf {h}}_{j}]$$ (Eq. 21)
The above equation not only applies to forward hidden state $\overrightarrow{\mathbf {h}}_{i}$ and backward hidden state $\overleftarrow{\mathbf {h}}_{i}$ , but also to the concatenation $\overleftrightarrow{\mathbf {h}}_{i}= [\overrightarrow{\mathbf {h}}_{i}, \overleftarrow{\mathbf {h}}_{i}]$ and summation $\mathbf {h}^{+}_{i}= \overrightarrow{\mathbf {h}}_{i} + \overleftarrow{\mathbf {h}}_{i}$ , resulting in a tensor $\mathbf {D}^{13 \times |sent1| \times |sent2|}$ after padding one extra bias term. A “hard” attention is applied to the interaction tensor to build word alignment: selecting the most related word pairs and increasing the corresponding weights by 10 times. Then a 19-layer deep CNN is applied to aggregate the word interaction features for final classification.
The Decomposable Attention Model (DecAtt)
The Decomposable Attention model BIBREF10 is one of the earliest models to introduce attention-based alignment for sentence pair modeling, and it achieved state-of-the-art results on the SNLI dataset with about an order of magnitude fewer parameters than other models (see more in Table 5 ) without relying on word order information. It computes the word pair interaction between $\mathbf {w}_{i}^{a}$ and $\mathbf {w}_{j}^{b}$ (from input sentences $s_a$ and $s_b$ , each with $m$ and $n$ words, respectively) as ${e}_{ij} = {F(\mathbf {w}_{i}^{a})}^{T} F(\mathbf {w}_{j}^{b})$ , where $F$ is a feedforward network; then alignment is determined as follows:
$$&\mathbf {\beta }_{i} = \sum _{j=1}^{n} \frac{exp({e}_{ij})}{\sum _{k=1}^{n} exp({e}_{ik})} \mathbf {w}_{j}^{b} \quad &\mathbf {\alpha }_{j} = \sum _{i=1}^{m} \frac{exp({e}_{ij})}{\sum _{k=1}^{m} exp({e}_{kj})} \mathbf {w}_{i}^{a} $$ (Eq. 23)
where $\mathbf {\beta }_{i}$ is the soft alignment between $\mathbf {w}_{i}^{a}$ and subphrases $\mathbf {w}_{j}^{b}$ in sentence $s_b$ , and vice versa for $\mathbf {\alpha }_{j}$ . The aligned phrases are fed into another feedforward network $G$ : $\mathbf {v}_{i}^{a} = G([\mathbf {w}_{i}^{a}; \mathbf {\beta }_{i}]) $ and $\mathbf {v}_{j}^{b} = G([\mathbf {w}_{j}^{b}; \mathbf {\alpha }_{j}]) $ to generate sets $\lbrace \mathbf {v}_{i}^{a}\rbrace $ and $\lbrace \mathbf {v}_{j}^{b}\rbrace $ , which are aggregated by summation and then concatenated together for classification.
The Enhanced Sequential Inference Model (ESIM)
The Enhanced Sequential Inference Model BIBREF0 is closely related to the DecAtt model, but it differs in a few aspects. First, Chen et al. Chen-Qian:2017:ACL demonstrated that using Bi-LSTM to encode sequential contexts is important for performance improvement. They used the concatenation $\overline{\mathbf {w}}_{i} = \overleftrightarrow{\mathbf {h}}_{i} = [\overrightarrow{\mathbf {h}}_{i}, \overleftarrow{\mathbf {h}}_{i}]$ of both directions as in the PWIM model. The word alignment $\mathbf {\beta }_{i}$ and $\mathbf {\alpha }_{j}$ between $\overline{\mathbf {w}}^{a}$ and $\overline{\mathbf {w}}^{b}$ are calculated the same way as in DecAtt. Second, they showed the competitive performance of recursive architecture with constituency parsing, which complements with sequential LSTM. The feedforward function $G$ in DecAtt is replaced with Tree-LSTM:
$$&\mathbf {v}_{i}^{a} = TreeLSTM([\overline{\mathbf {w}}_{i}^{a}; \mathbf {\beta }_{i}; \overline{\mathbf {w}}_{i}^{a}-\mathbf {\beta }_{i}; \overline{\mathbf {w}}_{i}^{a} \odot \mathbf {\beta }_{i}])\\ &\mathbf {v}_{j}^{b} = TreeLSTM([\overline{\mathbf {w}}_{j}^{b}; \mathbf {\alpha }_{j}; \overline{\mathbf {w}}_{j}^{b}-\mathbf {\alpha }_{j}; \overline{\mathbf {w}}_{j}^{b} \odot \mathbf {\alpha }_{j}])$$ (Eq. 25)
Third, instead of using summation in aggregation, ESIM adapts the average and max pooling and concatenation $\mathbf {v}= [\mathbf {v}_{ave}^{a}; \mathbf {v}_{max}^{a}; \mathbf {v}_{ave}^{b}; \mathbf {v}_{max}^{b}]$ before passing through multi-layer perceptron (MLP) for classification:
$$& \mathbf {v}_{ave}^{a} = \sum _{i=1}^{m} \frac{\mathbf {v}_{i}^{a}}{m}, \qquad \mathbf {v}_{max}^{a} = \max _{i=1}^{m}\mathbf {v}_{i}^{a} , \qquad \mathbf {v}_{ave}^{b} = \sum _{j=1}^{n} \frac{\mathbf {v}_{j}^{b}}{n}, \qquad \mathbf {v}_{max}^{b} = \max _{j=1}^{n}\mathbf {v}_{j}^{b}$$ (Eq. 26)
Datasets
We conducted sentence pair modeling experiments on eight popular datasets: two NLI datasets, three PI datasets, one STS dataset and two QA datasets. Table 2 gives a comparison of these datasets:
SNLI BIBREF7 contains 570k hypotheses written by crowdsourcing workers given the premises. It focuses on three semantic relations: the premise entails the hypothesis (entailment), they contradict each other (contradiction), or they are unrelated (neutral).
Multi-NLI BIBREF33 extends the SNLI corpus to multiple genres of written and spoken texts with 433k sentence pairs.
Quora BIBREF34 contains 400k question pairs collected from the Quora website. This dataset has balanced positive and negative labels indicating whether the questions are duplicated or not.
Twitter-URL BIBREF35 includes 50k sentence pairs collected from tweets that share the same URL of news articles. This dataset contains both formal and informal language.
PIT-2015 BIBREF5 comes from SemEval-2015 and was collected from tweets under the same trending topic. It contains naturally occurred (i.e. written by independent Twitter users spontaneously) paraphrases and non-paraphrases with varied topics and language styles.
STS-2014 BIBREF36 is from SemEval-2014, constructed from image descriptions, news headlines, tweet news, discussion forums, and OntoNotes BIBREF37 .
WikiQA BIBREF8 is an open-domain question-answering dataset. Following He and Lin he-lin:2016:N16-1, questions without correct candidate answer sentences are excluded, and answer sentences are truncated to 40 tokens, resulting in 12k question-answer pairs for our experiments.
TrecQA BIBREF38 is an answer selection task of 56k question-answer pairs and created in Text Retrieval Conferences (TREC). For both WikiQA and TrecQA datasets, the best answer is selected according to the semantic relatedness with the question.
Implementation Details
We implement all the models with the same PyTorch framework. Below, we summarize the implementation details that are key for reproducing results for each model:
SSE: This model can converge very fast, for example, 2 or 3 epochs for the SNLI dataset. We control the convergence speed by updating the learning rate for each epoch: specifically, $lr=\frac{1}{2^{\frac{epoch\_i}{2}}}*{init\_lr}$ , where $init\_lr$ is the initial learning rate and $epoch\_i$ is the index of current epoch.
DecAtt: It is important to use gradient clipping for this model: for each gradient update, we check the L2 norm of all the gradient values, if it is greater than a threshold $b$ , we scale the gradient by a factor $\alpha = b/L2\_norm$ . Another useful procedure is to assemble batches of sentences with similar length.
ESIM: Similar but different from DecAtt, ESIM batches sentences with varied length and uses masks to filter out padding information. In order to batch the parse trees within Tree-LSTM recursion, we follow Bowman et al.'s bowman-EtAl:2016:P16-1 procedure that converts tree structures into the linear sequential structure of a shift reduce parser. Two additional masks are used for producing left and right children of a tree node.
PWIM: The cosine and Euclidean distances used in the word interaction layer have smaller values for similar vectors while dot products have larger values. The performance increases if we add a negative sign to make all the vector similarity measurements behave consistently.
Analysis
Table 3 and 4 show the results reported in the original papers and the replicated results with our implementation. We use accuracy, F1 score, Pearson's $r$ , Mean Average Precision (MAP), and Mean Reciprocal Rank (MRR) for evaluation on different datasets following the literature. Our reproduced results are slightly lower than the original results by 0.5 $\sim $ 1.5 points on accuracy. We suspect the following potential reasons: (i) less extensive hyperparameter tuning for each individual dataset; (ii) only one run with random seeding to report results; and (iii) use of different neural network toolkits: for example, the original ESIM model was implemented with Theano, and PWIM model was in Torch.
Herein, we examine the main components that account for performance in sentence pair modeling.
How important is LSTM encoded context information for sentence pair modeling?
Regarding DecAtt, Parikh et al. parikh-EtAl:2016:EMNLP2016 mentioned that “intra-sentence attention is optional"; they can achieve competitive results without considering context information. However, not surprisingly, our experiments consistently show that encoding sequential context information with LSTM is critical. Compared to DecAtt, ESIM shows better performance on every dataset (see Table 4 and Figure 3 ). The main difference between ESIM and DecAtt that contributes to performance improvement, we found, is the use of Bi-LSTM and Tree-LSTM for sentence encoding, rather than the different choices of aggregation functions.
Why does Tree-LSTM help with Twitter data?
Chen et al. Chen-Qian:2017:ACL offered a simple combination (ESIM $_{seq+tree}$ ) by averaging the prediction probabilities of two ESIM variants that use sequential Bi-LSTM and Tree-LSTM respectively, and suggested “parsing information complements very well with ESIM and further improves the performance". However, we found that adding Tree-LSTM only helps slightly or not at all for most datasets, but it helps noticably with the two Twitter paraphrase datasets. We hypothesize the reason is that these two datasets come from real-world tweets which often contain extraneous text fragments, in contrast to SNLI and other datasets that have sentences written by crowdsourcing workers. For example, the segment “ever wondered ,” in the sentence pair ever wondered , why your recorded #voice sounds weird to you? and why do our recorded voices sound so weird to us? introduces a disruptive context into the Bi-LSTM encoder, while Tree-LSTM can put it in a less important position after constituency parsing.
How important is attentive interaction for sentence pair modeling? Why does SSE excel on Quora?
Both ESIM and DecAtt (Eq. 23 ) calculate an attention-based soft alignment between a sentence pair, which was also proposed in BIBREF27 and BIBREF29 for sentence pair modeling, whereas PWIM utilizes a hard attention mechanism. Both attention strategies are critical for model performance. In PWIM model BIBREF1 , we observed a 1 $\sim $ 2 point performance drop after removing the hard attention, 0 $\sim $ 3 point performance drop and $\sim $ 25% training time reduction after removing the 19-layer CNN aggregation. Likely without even the authors of SSE knowing, the SSE model performs extraordinarily well on the Quora corpus, perhaps because Quora contains many sentence pairs with less complicated inter-sentence interactions (e.g., many identical words in the two sentences) and incorrect ground truth labels (e.g., What is your biggest regret in life? and What's the biggest regret you've had in life? are labeled as non-duplicate questions by mistake).
Figure 3 shows the learning curves. The DecAtt model converges quickly and performs well on large NLI datasets due to its design simplicity. PWIM is the slowest model (see time comparison in Table 5 ) but shows very strong performance on semantic similarity and paraphrase identification datasets. ESIM and SSE keep a good balance between training time and performance.
[3]This number was reported in BIBREF12 by co-authors of DecAtt BIBREF10 . [4]This number was reproduced by Williams et al. williams2017broad. [5]This number was generated by InferSent traind on SNLI and Multi-NLI datasets.
As shown in Figure 4 , we experimented with different training sizes of the largest SNLI dataset. All the models show improved performance as we increase the training size. ESIM and SSE have very similar trends and clearly outperform PWIM on the SNLI dataset. DecAtt shows a performance jump when the training size exceeds a threshold.
We conducted an in-depth analysis of model performance on the Multi-domain NLI dataset based on different categories: text genre, sentence pair overlap, and sentence length. As shown in Table 6 , all models have comparable performance between matched genre and unmatched genre. Sentence length and overlap turn out to be two important factors – the longer the sentences and the fewer tokens in common, the more challenging it is to determine their semantic relationship. These phenomena shared by the state-of-the-art systems reflect their similar design framework which is symmetric at processing both sentences in the pair, while question answering and natural language inference tasks are directional BIBREF30 . How to incorporate asymmetry into model design will be worth more exploration in future research.
In addition to the cross-domain study (Table 6 ), we conducted transfer learning experiments on three paraphrase identification datasets (Table 4 ). The most noteworthy phenomenon is that the SSE model performs better on Twitter-URL and PIT-2015 when trained on the large out-of-domain Quora data than the small in-domain training data. Two likely reasons are: (i) the SSE model with over 29 million parameters is data hungry and (ii) SSE model is a sentence encoding model, which generalizes better across domains/tasks than sentence pair interaction models. Sentence pair interaction models may encounter difficulties on Quora, which contains sentence pairs with the highest word overlap (51.5%) among all datasets and often causes the interaction patterns to focus on a few key words that differ. In contrast, the Twitter-URL dataset has the lowest overlap (23.0%) with a semantic relationship that is mainly based on the intention of the tweets.
Conclusion
We analyzed five different neural models (and their variations) for sentence pair modeling and conducted a series of experiments with eight representative datasets for different NLP tasks. We quantified the importance of the LSTM encoder and attentive alignment for inter-sentence interaction, as well as the transfer learning ability of sentence encoding based models. We showed that the SNLI corpus of over 550k sentence pairs cannot saturate the learning curve. We systematically compared the strengths and weaknesses of different network designs and provided insights for future work.
Acknowledgements
We thank Ohio Supercomputer Center BIBREF39 for computing resources. This work was supported in part by NSF CRII award (RI-1755898) and DARPA through the ARO (W911NF-17-C-0095). The content of the information in this document does not necessarily reflect the position or the policy of the U.S. Government, and no official endorsement should be inferred.
Pretrained Word Embeddings
We used the 200-dimensional GloVe word vectors BIBREF18 , trained on 27 billion words from Twitter (vocabulary size of 1.2 milion words) for Twitter URL BIBREF35 and PIT-2015 BIBREF5 datasets, and the 300-dimensional GloVe vectors, trained on 840 billion words (vocabulary size of 2.2 milion words) from Common Crawl for all other datasets. For out-of-vocabulary words, we initialized the word vectors using normal distribution with mean 0 and deviation 1.
Hyper-parameter Settings
We followed original papers or code implementations to set hyper-parameters for these models. In Infersent model BIBREF16 , the hidden dimension size for Bi-LSTM is 2048, and the fully connected layers have 512 hidden units. In SSE model BIBREF17 , the hidden size for three Bi-LSTMs is 512, 2014 and 2048, respectively. The fully connected layers have 1600 units. PWIM BIBREF1 and ESIM BIBREF0 both use Bi-LSTM for context encoding, having 200 hidden units and 300 hidden units respectively. The DecAtt model BIBREF10 uses three kinds of feed forward networks, all of which have 300 hidden units. Other parameters like learning rate, batch size, dropout rate, and all of them use the same settings as in original papers.
Fine-tuning the Models
It is not practical to fine tune every hyper-parameter in every model and every dataset, since we want to show how these models can generalize well on other datasets, we need try to avoid fine-tuning these parameters on some specific datasets, otherwise we can easily get over-fitted models. Therefore, we keep the hyper-parameters unchanged across different datasets, to demonstrate the generalization capability of each model. The default number of epochs for training these models is set to 20, if some models could converge earlier (no more performance gain on development set), we would stop running them before they approached epoch 20. The 20 epochs can guarantee every model get converged on every dataset.
Question:
Do the authors also analyze transformer-based architectures?
Answer:
| No | 328 | 23,811 | 23,813 | 23,883 | ... [The rest of the article is omitted]
|
8bf7f1f93d0a2816234d36395ab40c481be9a0e0 | 8bf7f1f93d0a2816234d36395ab40c481be9a0e0_1 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
This work is licensed under a Creative Commons Attribution 4.0 International License. License details: http://creativecommons.org/licenses/by/4.0/.
Sentence pair modeling is a fundamental technique underlying many NLP tasks, including the following:
Traditionally, researchers had to develop different methods specific for each task. Now neural networks can perform all the above tasks with the same architecture by training end to end. Various neural models BIBREF1 , BIBREF0 , BIBREF10 , BIBREF11 , BIBREF12 , BIBREF13 , BIBREF14 , BIBREF15 have declared state-of-the-art results for sentence pair modeling tasks; however, they were carefully designed and evaluated on selected (often one or two) datasets that can demonstrate the superiority of the model. The research questions are as follows: Do they perform well on other tasks and datasets? How much performance gain is due to certain system design choices and hyperparameter optimizations?
To answer these questions and better understand different network designs, we systematically analyze and compare the state-of-the-art neural models across multiple tasks and multiple domains. Namely, we implement five models and their variations on the same PyTorch platform: InferSent model BIBREF16 , Shortcut-stacked Sentence Encoder Model BIBREF17 , Pairwise Word Interaction Model BIBREF1 , Decomposable Attention Model BIBREF10 , and Enhanced Sequential Inference Model BIBREF0 . They are representative of the two most common approaches: sentence encoding models that learn vector representations of individual sentences and then calculate the semantic relationship between sentences based on vector distance and sentence pair interaction models that use some sorts of word alignment mechanisms (e.g., attention) then aggregate inter-sentence interactions. We focus on identifying important network designs and present a series of findings with quantitative measurements and in-depth analyses, including (i) incorporating inter-sentence interactions is critical; (ii) Tree-LSTM does not help as much as previously claimed but surprisingly improves performance on Twitter data; (iii) Enhanced Sequential Inference Model has the most consistent high performance for larger datasets, while Pairwise Word Interaction Model performs better on smaller datasets and Shortcut-Stacked Sentence Encoder Model is the best performaning model on the Quora corpus. We release our implementations as a toolkit to the research community.
General Framework for Sentence Pair Modeling
Various neural networks have been proposed for sentence pair modeling, all of which fall into two types of approaches. The sentence encoding approach encodes each sentence into a fixed-length vector and then computes sentence similarity directly. The model of this type has advantages in the simplicity of the network design and generalization to other NLP tasks. The sentence pair interaction approach takes word alignment and interactions between the sentence pair into account and often show better performance when trained on in-domain data. Here we outline the two types of neural networks under the same general framework:
Representative Models for Sentence Pair Modeling
Table 1 gives a summary of typical models for sentence pair modeling in recent years. In particular, we investigate five models in depth: two are representative of the sentence encoding type of model, and three are representative of the interaction-aggregation type of model. These models have reported state-or-the-art results with varied architecture design (this section) and implementation details (Section "Implementation Details" ).
The Bi-LSTM Max-pooling Network (InferSent)
We choose the simple Bi-LSTM max-pooling network from InferSent BIBREF16 :
$$&\overleftrightarrow{\mathbf {h}}_{i} = BiLSTM(\mathbf {x}_{i}, \overleftrightarrow{\mathbf {h}}_{i-1}) \\ &\mathbf {v}=max(\overleftrightarrow{\mathbf {h}}_{1}, \overleftrightarrow{\mathbf {h}}_{2}, ..., \overleftrightarrow{\mathbf {h}}_{n})$$ (Eq. 17)
where $\overleftrightarrow{\mathbf {h}}_{i}$ represents the concatenation of hidden states in both directons. It has shown better transfer learning capabilities than several other sentence embedding models, including SkipThought BIBREF31 and FastSent BIBREF32 , when trained on the natural language inference datasets.
The Shortcut-Stacked Sentence Encoder Model (SSE)
The Shortcut-Stacked Sentence Encoder model BIBREF17 is a sentence-based embedding model, which enhances multi-layer Bi-LSTM with skip connection to avoid training error accumulation, and calculates each layer as follows:
$$&\overleftrightarrow{\mathbf {h}}_{i}^{k} = BiLSTM(\mathbf {x}_{i}^{k}, \overleftrightarrow{\mathbf {h}}_{i-1}^{k}) \\ &\mathbf {x}_{i}^{1}=\mathbf {w}_{i} \quad (k=1), \qquad \mathbf {x}_{i}^{k}=[\mathbf {w}_{i}, \overleftrightarrow{\mathbf {h}}_{i}^{k-1}, \overleftrightarrow{\mathbf {h}}_{i}^{k-2}, ..., \overleftrightarrow{\mathbf {h}}_{i}^{1}] \quad (k>1) \\ &\mathbf {v}=max(\overleftrightarrow{\mathbf {h}}_{1}^{m}, \overleftrightarrow{\mathbf {h}}_{2}^{m}, ..., \overleftrightarrow{\mathbf {h}}_{n}^{m})$$ (Eq. 19)
where $\mathbf {x}_{i}^{k}$ is the input of the $k$ th Bi-LSTM layer at time step $i$ , which is the combination of outputs from all previous layers, $\overleftrightarrow{\mathbf {h}}_{i}^{k}$ represents the hidden state of the $k$ th Bi-LSTM layer in both directions. The final sentence embedding $\mathbf {v}$ is the row-based max pooling over the output of the last Bi-LSTM layer, where $n$ denotes the number of words within a sentence and $m$ is the number of Bi-LSTM layers ( $m = 3$ in SSE).
The Pairwise Word Interaction Model (PWIM)
In the Pairwise Word Interaction model BIBREF1 , each word vector $\mathbf {w}_{i}$ is encoded with context through forward and backward LSTMs: $\overrightarrow{\mathbf {h}}_{i} = LSTM^{f}(\mathbf {w}_{i}, \overrightarrow{\mathbf {h}}_{i-1})$ and $\overleftarrow{\mathbf {h}}_{i} = LSTM^{b}(\mathbf {w}_{i}, \overleftarrow{\mathbf {h}}_{i+1})$ . For every word pair $(\mathbf {w}^a_{i}, \mathbf {w}^b_{j})$ across sentences, the model directly calculates word pair interactions using cosine similarity, Euclidean distance, and dot product over the outputs of the encoding layer:
$$D(\overrightarrow{\mathbf {h}}_{i}, \overrightarrow{\mathbf {h}}_{j}) & = [cos(\overrightarrow{\mathbf {h}}_{i}, \overrightarrow{\mathbf {h}}_{j}), \Vert \overrightarrow{\mathbf {h}}_{i} - \overrightarrow{\mathbf {h}}_{j}\Vert , \overrightarrow{\mathbf {h}}_{i} \cdot \overrightarrow{\mathbf {h}}_{j}]$$ (Eq. 21)
The above equation not only applies to forward hidden state $\overrightarrow{\mathbf {h}}_{i}$ and backward hidden state $\overleftarrow{\mathbf {h}}_{i}$ , but also to the concatenation $\overleftrightarrow{\mathbf {h}}_{i}= [\overrightarrow{\mathbf {h}}_{i}, \overleftarrow{\mathbf {h}}_{i}]$ and summation $\mathbf {h}^{+}_{i}= \overrightarrow{\mathbf {h}}_{i} + \overleftarrow{\mathbf {h}}_{i}$ , resulting in a tensor $\mathbf {D}^{13 \times |sent1| \times |sent2|}$ after padding one extra bias term. A “hard” attention is applied to the interaction tensor to build word alignment: selecting the most related word pairs and increasing the corresponding weights by 10 times. Then a 19-layer deep CNN is applied to aggregate the word interaction features for final classification.
The Decomposable Attention Model (DecAtt)
The Decomposable Attention model BIBREF10 is one of the earliest models to introduce attention-based alignment for sentence pair modeling, and it achieved state-of-the-art results on the SNLI dataset with about an order of magnitude fewer parameters than other models (see more in Table 5 ) without relying on word order information. It computes the word pair interaction between $\mathbf {w}_{i}^{a}$ and $\mathbf {w}_{j}^{b}$ (from input sentences $s_a$ and $s_b$ , each with $m$ and $n$ words, respectively) as ${e}_{ij} = {F(\mathbf {w}_{i}^{a})}^{T} F(\mathbf {w}_{j}^{b})$ , where $F$ is a feedforward network; then alignment is determined as follows:
$$&\mathbf {\beta }_{i} = \sum _{j=1}^{n} \frac{exp({e}_{ij})}{\sum _{k=1}^{n} exp({e}_{ik})} \mathbf {w}_{j}^{b} \quad &\mathbf {\alpha }_{j} = \sum _{i=1}^{m} \frac{exp({e}_{ij})}{\sum _{k=1}^{m} exp({e}_{kj})} \mathbf {w}_{i}^{a} $$ (Eq. 23)
where $\mathbf {\beta }_{i}$ is the soft alignment between $\mathbf {w}_{i}^{a}$ and subphrases $\mathbf {w}_{j}^{b}$ in sentence $s_b$ , and vice versa for $\mathbf {\alpha }_{j}$ . The aligned phrases are fed into another feedforward network $G$ : $\mathbf {v}_{i}^{a} = G([\mathbf {w}_{i}^{a}; \mathbf {\beta }_{i}]) $ and $\mathbf {v}_{j}^{b} = G([\mathbf {w}_{j}^{b}; \mathbf {\alpha }_{j}]) $ to generate sets $\lbrace \mathbf {v}_{i}^{a}\rbrace $ and $\lbrace \mathbf {v}_{j}^{b}\rbrace $ , which are aggregated by summation and then concatenated together for classification.
The Enhanced Sequential Inference Model (ESIM)
The Enhanced Sequential Inference Model BIBREF0 is closely related to the DecAtt model, but it differs in a few aspects. First, Chen et al. Chen-Qian:2017:ACL demonstrated that using Bi-LSTM to encode sequential contexts is important for performance improvement. They used the concatenation $\overline{\mathbf {w}}_{i} = \overleftrightarrow{\mathbf {h}}_{i} = [\overrightarrow{\mathbf {h}}_{i}, \overleftarrow{\mathbf {h}}_{i}]$ of both directions as in the PWIM model. The word alignment $\mathbf {\beta }_{i}$ and $\mathbf {\alpha }_{j}$ between $\overline{\mathbf {w}}^{a}$ and $\overline{\mathbf {w}}^{b}$ are calculated the same way as in DecAtt. Second, they showed the competitive performance of recursive architecture with constituency parsing, which complements with sequential LSTM. The feedforward function $G$ in DecAtt is replaced with Tree-LSTM:
$$&\mathbf {v}_{i}^{a} = TreeLSTM([\overline{\mathbf {w}}_{i}^{a}; \mathbf {\beta }_{i}; \overline{\mathbf {w}}_{i}^{a}-\mathbf {\beta }_{i}; \overline{\mathbf {w}}_{i}^{a} \odot \mathbf {\beta }_{i}])\\ &\mathbf {v}_{j}^{b} = TreeLSTM([\overline{\mathbf {w}}_{j}^{b}; \mathbf {\alpha }_{j}; \overline{\mathbf {w}}_{j}^{b}-\mathbf {\alpha }_{j}; \overline{\mathbf {w}}_{j}^{b} \odot \mathbf {\alpha }_{j}])$$ (Eq. 25)
Third, instead of using summation in aggregation, ESIM adapts the average and max pooling and concatenation $\mathbf {v}= [\mathbf {v}_{ave}^{a}; \mathbf {v}_{max}^{a}; \mathbf {v}_{ave}^{b}; \mathbf {v}_{max}^{b}]$ before passing through multi-layer perceptron (MLP) for classification:
$$& \mathbf {v}_{ave}^{a} = \sum _{i=1}^{m} \frac{\mathbf {v}_{i}^{a}}{m}, \qquad \mathbf {v}_{max}^{a} = \max _{i=1}^{m}\mathbf {v}_{i}^{a} , \qquad \mathbf {v}_{ave}^{b} = \sum _{j=1}^{n} \frac{\mathbf {v}_{j}^{b}}{n}, \qquad \mathbf {v}_{max}^{b} = \max _{j=1}^{n}\mathbf {v}_{j}^{b}$$ (Eq. 26)
Datasets
We conducted sentence pair modeling experiments on eight popular datasets: two NLI datasets, three PI datasets, one STS dataset and two QA datasets. Table 2 gives a comparison of these datasets:
SNLI BIBREF7 contains 570k hypotheses written by crowdsourcing workers given the premises. It focuses on three semantic relations: the premise entails the hypothesis (entailment), they contradict each other (contradiction), or they are unrelated (neutral).
Multi-NLI BIBREF33 extends the SNLI corpus to multiple genres of written and spoken texts with 433k sentence pairs.
Quora BIBREF34 contains 400k question pairs collected from the Quora website. This dataset has balanced positive and negative labels indicating whether the questions are duplicated or not.
Twitter-URL BIBREF35 includes 50k sentence pairs collected from tweets that share the same URL of news articles. This dataset contains both formal and informal language.
PIT-2015 BIBREF5 comes from SemEval-2015 and was collected from tweets under the same trending topic. It contains naturally occurred (i.e. written by independent Twitter users spontaneously) paraphrases and non-paraphrases with varied topics and language styles.
STS-2014 BIBREF36 is from SemEval-2014, constructed from image descriptions, news headlines, tweet news, discussion forums, and OntoNotes BIBREF37 .
WikiQA BIBREF8 is an open-domain question-answering dataset. Following He and Lin he-lin:2016:N16-1, questions without correct candidate answer sentences are excluded, and answer sentences are truncated to 40 tokens, resulting in 12k question-answer pairs for our experiments.
TrecQA BIBREF38 is an answer selection task of 56k question-answer pairs and created in Text Retrieval Conferences (TREC). For both WikiQA and TrecQA datasets, the best answer is selected according to the semantic relatedness with the question.
Implementation Details
We implement all the models with the same PyTorch framework. Below, we summarize the implementation details that are key for reproducing results for each model:
SSE: This model can converge very fast, for example, 2 or 3 epochs for the SNLI dataset. We control the convergence speed by updating the learning rate for each epoch: specifically, $lr=\frac{1}{2^{\frac{epoch\_i}{2}}}*{init\_lr}$ , where $init\_lr$ is the initial learning rate and $epoch\_i$ is the index of current epoch.
DecAtt: It is important to use gradient clipping for this model: for each gradient update, we check the L2 norm of all the gradient values, if it is greater than a threshold $b$ , we scale the gradient by a factor $\alpha = b/L2\_norm$ . Another useful procedure is to assemble batches of sentences with similar length.
ESIM: Similar but different from DecAtt, ESIM batches sentences with varied length and uses masks to filter out padding information. In order to batch the parse trees within Tree-LSTM recursion, we follow Bowman et al.'s bowman-EtAl:2016:P16-1 procedure that converts tree structures into the linear sequential structure of a shift reduce parser. Two additional masks are used for producing left and right children of a tree node.
PWIM: The cosine and Euclidean distances used in the word interaction layer have smaller values for similar vectors while dot products have larger values. The performance increases if we add a negative sign to make all the vector similarity measurements behave consistently.
Analysis
Table 3 and 4 show the results reported in the original papers and the replicated results with our implementation. We use accuracy, F1 score, Pearson's $r$ , Mean Average Precision (MAP), and Mean Reciprocal Rank (MRR) for evaluation on different datasets following the literature. Our reproduced results are slightly lower than the original results by 0.5 $\sim $ 1.5 points on accuracy. We suspect the following potential reasons: (i) less extensive hyperparameter tuning for each individual dataset; (ii) only one run with random seeding to report results; and (iii) use of different neural network toolkits: for example, the original ESIM model was implemented with Theano, and PWIM model was in Torch.
Herein, we examine the main components that account for performance in sentence pair modeling.
How important is LSTM encoded context information for sentence pair modeling?
Regarding DecAtt, Parikh et al. parikh-EtAl:2016:EMNLP2016 mentioned that “intra-sentence attention is optional"; they can achieve competitive results without considering context information. However, not surprisingly, our experiments consistently show that encoding sequential context information with LSTM is critical. Compared to DecAtt, ESIM shows better performance on every dataset (see Table 4 and Figure 3 ). The main difference between ESIM and DecAtt that contributes to performance improvement, we found, is the use of Bi-LSTM and Tree-LSTM for sentence encoding, rather than the different choices of aggregation functions.
Why does Tree-LSTM help with Twitter data?
Chen et al. Chen-Qian:2017:ACL offered a simple combination (ESIM $_{seq+tree}$ ) by averaging the prediction probabilities of two ESIM variants that use sequential Bi-LSTM and Tree-LSTM respectively, and suggested “parsing information complements very well with ESIM and further improves the performance". However, we found that adding Tree-LSTM only helps slightly or not at all for most datasets, but it helps noticably with the two Twitter paraphrase datasets. We hypothesize the reason is that these two datasets come from real-world tweets which often contain extraneous text fragments, in contrast to SNLI and other datasets that have sentences written by crowdsourcing workers. For example, the segment “ever wondered ,” in the sentence pair ever wondered , why your recorded #voice sounds weird to you? and why do our recorded voices sound so weird to us? introduces a disruptive context into the Bi-LSTM encoder, while Tree-LSTM can put it in a less important position after constituency parsing.
How important is attentive interaction for sentence pair modeling? Why does SSE excel on Quora?
Both ESIM and DecAtt (Eq. 23 ) calculate an attention-based soft alignment between a sentence pair, which was also proposed in BIBREF27 and BIBREF29 for sentence pair modeling, whereas PWIM utilizes a hard attention mechanism. Both attention strategies are critical for model performance. In PWIM model BIBREF1 , we observed a 1 $\sim $ 2 point performance drop after removing the hard attention, 0 $\sim $ 3 point performance drop and $\sim $ 25% training time reduction after removing the 19-layer CNN aggregation. Likely without even the authors of SSE knowing, the SSE model performs extraordinarily well on the Quora corpus, perhaps because Quora contains many sentence pairs with less complicated inter-sentence interactions (e.g., many identical words in the two sentences) and incorrect ground truth labels (e.g., What is your biggest regret in life? and What's the biggest regret you've had in life? are labeled as non-duplicate questions by mistake).
Figure 3 shows the learning curves. The DecAtt model converges quickly and performs well on large NLI datasets due to its design simplicity. PWIM is the slowest model (see time comparison in Table 5 ) but shows very strong performance on semantic similarity and paraphrase identification datasets. ESIM and SSE keep a good balance between training time and performance.
[3]This number was reported in BIBREF12 by co-authors of DecAtt BIBREF10 . [4]This number was reproduced by Williams et al. williams2017broad. [5]This number was generated by InferSent traind on SNLI and Multi-NLI datasets.
As shown in Figure 4 , we experimented with different training sizes of the largest SNLI dataset. All the models show improved performance as we increase the training size. ESIM and SSE have very similar trends and clearly outperform PWIM on the SNLI dataset. DecAtt shows a performance jump when the training size exceeds a threshold.
We conducted an in-depth analysis of model performance on the Multi-domain NLI dataset based on different categories: text genre, sentence pair overlap, and sentence length. As shown in Table 6 , all models have comparable performance between matched genre and unmatched genre. Sentence length and overlap turn out to be two important factors – the longer the sentences and the fewer tokens in common, the more challenging it is to determine their semantic relationship. These phenomena shared by the state-of-the-art systems reflect their similar design framework which is symmetric at processing both sentences in the pair, while question answering and natural language inference tasks are directional BIBREF30 . How to incorporate asymmetry into model design will be worth more exploration in future research.
In addition to the cross-domain study (Table 6 ), we conducted transfer learning experiments on three paraphrase identification datasets (Table 4 ). The most noteworthy phenomenon is that the SSE model performs better on Twitter-URL and PIT-2015 when trained on the large out-of-domain Quora data than the small in-domain training data. Two likely reasons are: (i) the SSE model with over 29 million parameters is data hungry and (ii) SSE model is a sentence encoding model, which generalizes better across domains/tasks than sentence pair interaction models. Sentence pair interaction models may encounter difficulties on Quora, which contains sentence pairs with the highest word overlap (51.5%) among all datasets and often causes the interaction patterns to focus on a few key words that differ. In contrast, the Twitter-URL dataset has the lowest overlap (23.0%) with a semantic relationship that is mainly based on the intention of the tweets.
Conclusion
We analyzed five different neural models (and their variations) for sentence pair modeling and conducted a series of experiments with eight representative datasets for different NLP tasks. We quantified the importance of the LSTM encoder and attentive alignment for inter-sentence interaction, as well as the transfer learning ability of sentence encoding based models. We showed that the SNLI corpus of over 550k sentence pairs cannot saturate the learning curve. We systematically compared the strengths and weaknesses of different network designs and provided insights for future work.
Acknowledgements
We thank Ohio Supercomputer Center BIBREF39 for computing resources. This work was supported in part by NSF CRII award (RI-1755898) and DARPA through the ARO (W911NF-17-C-0095). The content of the information in this document does not necessarily reflect the position or the policy of the U.S. Government, and no official endorsement should be inferred.
Pretrained Word Embeddings
We used the 200-dimensional GloVe word vectors BIBREF18 , trained on 27 billion words from Twitter (vocabulary size of 1.2 milion words) for Twitter URL BIBREF35 and PIT-2015 BIBREF5 datasets, and the 300-dimensional GloVe vectors, trained on 840 billion words (vocabulary size of 2.2 milion words) from Common Crawl for all other datasets. For out-of-vocabulary words, we initialized the word vectors using normal distribution with mean 0 and deviation 1.
Hyper-parameter Settings
We followed original papers or code implementations to set hyper-parameters for these models. In Infersent model BIBREF16 , the hidden dimension size for Bi-LSTM is 2048, and the fully connected layers have 512 hidden units. In SSE model BIBREF17 , the hidden size for three Bi-LSTMs is 512, 2014 and 2048, respectively. The fully connected layers have 1600 units. PWIM BIBREF1 and ESIM BIBREF0 both use Bi-LSTM for context encoding, having 200 hidden units and 300 hidden units respectively. The DecAtt model BIBREF10 uses three kinds of feed forward networks, all of which have 300 hidden units. Other parameters like learning rate, batch size, dropout rate, and all of them use the same settings as in original papers.
Fine-tuning the Models
It is not practical to fine tune every hyper-parameter in every model and every dataset, since we want to show how these models can generalize well on other datasets, we need try to avoid fine-tuning these parameters on some specific datasets, otherwise we can easily get over-fitted models. Therefore, we keep the hyper-parameters unchanged across different datasets, to demonstrate the generalization capability of each model. The default number of epochs for training these models is set to 20, if some models could converge earlier (no more performance gain on development set), we would stop running them before they approached epoch 20. The 20 epochs can guarantee every model get converged on every dataset.
Question:
Do the authors also analyze transformer-based architectures?
Answer:
| No | 328 | 23,811 | 23,813 | 23,883 | ... [The rest of the article is omitted]
|
0f12dc077fe8e5b95ca9163cea1dd17195c96929 | 0f12dc077fe8e5b95ca9163cea1dd17195c96929_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
[0]leftmargin=* [0]leftmargin=*
Automatic systems have had a significant and beneficial impact on all walks of human life. So much so that it is easy to overlook their potential to benefit society by promoting equity, diversity, and fairness. For example, machines do not take bribes to do their jobs, they can determine eligibility for a loan without being influenced by the color of the applicant's skin, and they can provide access to information and services without discrimination based on gender or sexual orientation. Nonetheless, as machine learning systems become more human-like in their predictions, they can also perpetuate human biases. Some learned biases may be beneficial for the downstream application (e.g., learning that humans often use some insect names, such as spider or cockroach, to refer to unpleasant situations). Other biases can be inappropriate and result in negative experiences for some groups of people. Examples include, loan eligibility and crime recidivism prediction systems that negatively assess people belonging to a certain pin/zip code (which may disproportionately impact people of a certain race) BIBREF0 and resumé sorting systems that believe that men are more qualified to be programmers than women BIBREF1 . Similarly, sentiment and emotion analysis systems can also perpetuate and accentuate inappropriate human biases, e.g., systems that consider utterances from one race or gender to be less positive simply because of their race or gender, or customer support systems that prioritize a call from an angry male over a call from the equally angry female.
Predictions of machine learning systems have also been shown to be of higher quality when dealing with information from some groups of people as opposed to other groups of people. For example, in the area of computer vision, gender classification systems perform particularly poorly for darker skinned females BIBREF2 . Natural language processing (NLP) systems have been shown to be poor in understanding text produced by people belonging to certain races BIBREF3 , BIBREF4 . For NLP systems, the sources of the bias often include the training data, other corpora, lexicons, and word embeddings that the machine learning algorithm may leverage to build its prediction model.
Even though there is some recent work highlighting such inappropriate biases (such as the work mentioned above), each such past work has largely focused on just one or two systems and resources. Further, there is no benchmark dataset for examining inappropriate biases in natural language systems. In this paper, we describe how we compiled a dataset of 8,640 English sentences carefully chosen to tease out biases towards certain races and genders. We will refer to it as the Equity Evaluation Corpus (EEC). We used the EEC as a supplementary test set in a recent shared task on predicting sentiment and emotion intensity in tweets, SemEval-2018 Task 1: Affect in Tweets BIBREF5 . In particular, we wanted to test a hypothesis that a system should equally rate the intensity of the emotion expressed by two sentences that differ only in the gender/race of a person mentioned. Note that here the term system refers to the combination of a machine learning architecture trained on a labeled dataset, and possibly using additional language resources. The bias can originate from any or several of these parts. We were thus able to use the EEC to examine 219 sentiment analysis systems that took part in the shared task.
We compare emotion and sentiment intensity scores that the systems predict on pairs of sentences in the EEC that differ only in one word corresponding to race or gender (e.g., `This man made me feel angry' vs. `This woman made me feel angry'). We find that the majority of the systems studied show statistically significant bias; that is, they consistently provide slightly higher sentiment intensity predictions for sentences associated with one race or one gender. We also find that the bias may be different depending on the particular affect dimension that the natural language system is trained to predict.
Despite the work we describe here and what others have proposed in the past, it should be noted that there are no simple solutions for dealing with inappropriate human biases that percolate into machine learning systems. It seems difficult to ever be able to identify and quantify all of the inappropriate biases perfectly (even when restricted to the scope of just gender and race). Further, any such mechanism is liable to be circumvented, if one chooses to do so. Nonetheless, as developers of sentiment analysis systems, and NLP systems more broadly, we cannot absolve ourselves of the ethical implications of the systems we build. Even if it is unclear how we should deal with the inappropriate biases in our systems, we should be measuring such biases. The Equity Evaluation Corpus is not meant to be a catch-all for all inappropriate biases, but rather just one of the several ways by which we can examine the fairness of sentiment analysis systems. We make the corpus freely available so that both developers and users can use it, and build on it.
Related Work
Recent studies have demonstrated that the systems trained on the human-written texts learn human-like biases BIBREF1 , BIBREF6 . In general, any predictive model built on historical data may inadvertently inherit human biases based on gender, ethnicity, race, or religion BIBREF7 , BIBREF8 . Discrimination-aware data mining focuses on measuring discrimination in data as well as on evaluating performance of discrimination-aware predictive models BIBREF9 , BIBREF10 , BIBREF11 , BIBREF12 .
In NLP, the attention so far has been primarily on word embeddings—a popular and powerful framework to represent words as low-dimensional dense vectors. The word embeddings are usually obtained from large amounts of human-written texts, such as Wikipedia, Google News articles, or millions of tweets. Bias in sentiment analysis systems has only been explored in simple systems that make use of pre-computed word embeddings BIBREF13 . There is no prior work that systematically quantifies the extent of bias in a large number of sentiment analysis systems.
This paper does not examine the differences in accuracies of systems on text produced by different races or genders, as was done by hovy2015demographic,blodgett2016demographic,jurgens2017incorporating,buolamwini2018gender. Approaches on how to mitigate inappropriate biases BIBREF14 , BIBREF1 , BIBREF15 , BIBREF16 , BIBREF13 , BIBREF17 , BIBREF18 are also beyond the scope of this paper. See also the position paper by hovy2016social, which identifies socio-ethical implications of the NLP systems in general.
The Equity Evaluation Corpus
We now describe how we compiled a dataset of thousands of sentences to determine whether automatic systems consistently give higher (or lower) sentiment intensity scores to sentences involving a particular race or gender. There are several ways in which such a dataset may be compiled. We present below the choices that we made.
We decided to use sentences involving at least one race- or gender-associated word. The sentences were intended to be short and grammatically simple. We also wanted some sentences to include expressions of sentiment and emotion, since the goal is to test sentiment and emotion systems. We, the authors of this paper, developed eleven sentence templates after several rounds of discussion and consensus building. They are shown in Table TABREF3 . The templates are divided into two groups. The first type (templates 1–7) includes emotion words. The purpose of this set is to have sentences expressing emotions. The second type (templates 8–11) does not include any emotion words. The purpose of this set is to have non-emotional (neutral) sentences.
The templates include two variables: INLINEFORM0 person INLINEFORM1 and INLINEFORM2 emotion word INLINEFORM3 . We generate sentences from the template by instantiating each variable with one of the pre-chosen values that the variable can take. Each of the eleven templates includes the variable INLINEFORM4 person INLINEFORM5 . INLINEFORM6 person INLINEFORM7 can be instantiated by any of the following noun phrases:
For our study, we chose ten names of each kind from the study by Caliskan:2017 (see Table TABREF7 ). The full lists of noun phrases representing females and males, used in our study, are shown in Table TABREF8 .
The second variable, INLINEFORM0 emotion word INLINEFORM1 , has two variants. Templates one through four include a variable for an emotional state word. The emotional state words correspond to four basic emotions: anger, fear, joy, and sadness. Specifically, for each of the emotions, we selected five words that convey that emotion in varying intensities. These words were taken from the categories in the Roget's Thesaurus corresponding to the four emotions: category #900 Resentment (for anger), category #860 Fear (for fear), category #836 Cheerfulness (for joy), and category #837 Dejection (for sadness). Templates five through seven include emotion words describing a situation or event. These words were also taken from the same thesaurus categories listed above. The full lists of emotion words (emotional state words and emotional situation/event words) are shown in Table TABREF10 .
We generated sentences from the templates by replacing INLINEFORM0 person INLINEFORM1 and INLINEFORM2 emotion word INLINEFORM3 variables with the values they can take. In total, 8,640 sentences were generated with the various combinations of INLINEFORM4 person INLINEFORM5 and INLINEFORM6 emotion word INLINEFORM7 values across the eleven templates. We manually examined the sentences to make sure they were grammatically well-formed. Notably, one can derive pairs of sentences from the EEC such that they differ only in one word corresponding to gender or race (e.g., `My daughter feels devastated' and `My son feels devastated'). We refer to the full set of 8,640 sentences as Equity Evaluation Corpus.
Measuring Race and Gender Bias in Automatic Sentiment Analysis Systems
The race and gender bias evaluation was carried out on the output of the 219 automatic systems that participated in SemEval-2018 Task 1: Affect in Tweets BIBREF5 . The shared task included five subtasks on inferring the affectual state of a person from their tweet: 1. emotion intensity regression, 2. emotion intensity ordinal classification, 3. valence (sentiment) regression, 4. valence ordinal classification, and 5. emotion classification. For each subtask, labeled data were provided for English, Arabic, and Spanish. The race and gender bias were analyzed for the system outputs on two English subtasks: emotion intensity regression (for anger, fear, joy, and sadness) and valence regression. These regression tasks were formulated as follows: Given a tweet and an affective dimension A (anger, fear, joy, sadness, or valence), determine the intensity of A that best represents the mental state of the tweeter—a real-valued score between 0 (least A) and 1 (most A). Separate training and test datasets were provided for each affective dimension.
Training sets included tweets along with gold intensity scores. Two test sets were provided for each task: 1. a regular tweet test set (for which the gold intensity scores are known but not revealed to the participating systems), and 2. the Equity Evaluation Corpus (for which no gold intensity labels exist). Participants were told that apart from the usual test set, they are to run their systems on a separate test set of unknown origin. The participants were instructed to train their system on the tweets training sets provided, and that they could use any other resources they may find or create. They were to run the same final system on the two test sets. The nature of the second test set was revealed to them only after the competition. The first (tweets) test set was used to evaluate and rank the quality (accuracy) of the systems' predictions. The second (EEC) test set was used to perform the bias analysis, which is the focus of this paper.
Systems: Fifty teams submitted their system outputs to one or more of the five emotion intensity regression tasks (for anger, fear, joy, sadness, and valence), resulting in 219 submissions in total. Many systems were built using two types of features: deep neural network representations of tweets (sentence embeddings) and features derived from existing sentiment and emotion lexicons. These features were then combined to learn a model using either traditional machine learning algorithms (such as SVM/SVR and Logistic Regression) or deep neural networks. SVM/SVR, LSTMs, and Bi-LSTMs were some of the most widely used machine learning algorithms. The sentence embeddings were obtained by training a neural network on the provided training data, a distant supervision corpus (e.g., AIT2018 Distant Supervision Corpus that has tweets with emotion-related query terms), sentiment-labeled tweet corpora (e.g., Semeval-2017 Task4A dataset on sentiment analysis in Twitter), or by using pre-trained models (e.g., DeepMoji BIBREF20 , Skip thoughts BIBREF21 ). The lexicon features were often derived from the NRC emotion and sentiment lexicons BIBREF22 , BIBREF23 , BIBREF24 , AFINN BIBREF25 , and Bing Liu Lexicon BIBREF26 .
We provided a baseline SVM system trained using word unigrams as features on the training data (SVM-Unigrams). This system is also included in the current analysis.
Measuring bias: To examine gender bias, we compared each system's predicted scores on the EEC sentence pairs as follows:
Thus, eleven pairs of scores (ten pairs of scores from ten noun phrase pairs and one pair of scores from the averages on name subsets) were examined for each template–emotion word instantiation. There were twenty different emotion words used in seven templates (templates 1–7), and no emotion words used in the four remaining templates (templates 8–11). In total, INLINEFORM0 pairs of scores were compared.
Similarly, to examine race bias, we compared pairs of system predicted scores as follows:
Thus, one pair of scores was examined for each template–emotion word instantiation. In total, INLINEFORM0 pairs of scores were compared.
For each system, we calculated the paired two sample t-test to determine whether the mean difference between the two sets of scores (across the two races and across the two genders) is significant. We set the significance level to 0.05. However, since we performed 438 assessments (219 submissions evaluated for biases in both gender and race), we applied Bonferroni correction. The null hypothesis that the true mean difference between the paired samples was zero was rejected if the calculated p-value fell below INLINEFORM0 .
Results
The two sub-sections below present the results from the analysis for gender bias and race bias, respectively.
Gender Bias Results
Individual submission results were communicated to the participants. Here, we present the summary results across all the teams. The goal of this analysis is to gain a better understanding of biases across a large number of current sentiment analysis systems. Thus, we partition the submissions into three groups according to the bias they show:
F=M not significant: submissions that showed no statistically significant difference in intensity scores predicted for corresponding female and male noun phrase sentences,
F INLINEFORM0 –M INLINEFORM1 significant: submissions that consistently gave higher scores for sentences with female noun phrases than for corresponding sentences with male noun phrases,
F INLINEFORM0 –M INLINEFORM1 significant: submissions that consistently gave lower scores for sentences with female noun phrases than for corresponding sentences with male noun phrases.
For each system and each sentence pair, we calculate the score difference INLINEFORM0 as the score for the female noun phrase sentence minus the score for the corresponding male noun phrase sentence. Table TABREF24 presents the summary results for each of the bias groups. It has the following columns:
#Subm.: number of submissions in each group.
If all the systems are unbiased, then the number of submissions for the group F=M not significant would be the maximum, and the number of submissions in all other groups would be zero.
Avg. score difference F INLINEFORM0 –M INLINEFORM1 : the average INLINEFORM2 for only those pairs where the score for the female noun phrase sentence is higher. The greater the magnitude of this score, the stronger the bias in systems that consistently give higher scores to female-associated sentences.
Avg. score difference F INLINEFORM0 –M INLINEFORM1 : the average INLINEFORM2 for only those pairs where the score for the female noun phrase sentence is lower. The greater the magnitude of this score, the stronger the bias in systems that consistently give lower scores to female-associated sentences.
Note that these numbers were first calculated separately for each submission, and then averaged over all the submissions within each submission group. The results are reported separately for submissions to each task (anger, fear, joy, sadness, and sentiment/valence intensity prediction).
Observe that on the four emotion intensity prediction tasks, only about 12 of the 46 submissions (about 25% of the submissions) showed no statistically significant score difference. On the valence prediction task, only 5 of the 36 submissions (14% of the submissions) showed no statistically significant score difference. Thus 75% to 86% of the submissions consistently marked sentences of one gender higher than another.
When predicting anger, joy, or valence, the number of systems consistently giving higher scores to sentences with female noun phrases (21–25) is markedly higher than the number of systems giving higher scores to sentences with male noun phrases (8–13). (Recall that higher valence means more positive sentiment.) In contrast, on the fear task, most submissions tended to assign higher scores to sentences with male noun phrases (23) as compared to the number of systems giving higher scores to sentences with female noun phrases (12). When predicting sadness, the number of submissions that mostly assigned higher scores to sentences with female noun phrases (18) is close to the number of submissions that mostly assigned higher scores to sentences with male noun phrases (16). These results are in line with some common stereotypes, such as females are more emotional, and situations involving male agents are more fearful BIBREF27 .
Figure FIGREF25 shows the score differences ( INLINEFORM0 ) for individual systems on the valence regression task. Plots for the four emotion intensity prediction tasks are shown in Figure FIGREF31 in the Appendix. Each point ( ▲, ▼, ●) on the plot corresponds to the difference in scores predicted by the system on one sentence pair. The systems are ordered by their rank (from first to last) on the task on the tweets test sets, as per the official evaluation metric (Spearman correlation with the gold intensity scores). We will refer to the difference between the maximal value of INLINEFORM1 and the minimal value of INLINEFORM2 for a particular system as the INLINEFORM3 –spread. Observe that the INLINEFORM4 –spreads for many systems are rather large, up to 0.57. Depending on the task, the top 10 or top 15 systems as well as some of the worst performing systems tend to have smaller INLINEFORM5 –spreads while the systems with medium to low performance show greater sensitivity to the gender-associated words. Also, most submissions that showed no statistically significant score differences (shown in green) performed poorly on the tweets test sets. Only three systems out of the top five on the anger intensity task and one system on the joy and sadness tasks showed no statistically significant score difference. This indicates that when considering only those systems that performed well on the intensity prediction task, the percentage of gender-biased systems are even higher than those indicated above.
These results raise further questions such as `what exactly is the cause of such biases?' and `why is the bias impacted by the emotion task under consideration?'. Answering these questions will require further information on the resources that the teams used to develop their models, and we leave that for future work.
Average score differences: For submissions that showed statistically significant score differences, the average score difference F INLINEFORM0 –M INLINEFORM1 and the average score difference F INLINEFORM2 –M INLINEFORM3 were INLINEFORM4 . Since the intensity scores range from 0 to 1, 0.03 is 3% of the full range. The maximal score difference ( INLINEFORM5 ) across all the submissions was as high as 0.34. Note, however, that these INLINEFORM6 s are the result of changing just one word in a sentence. In more complex sentences, several gender-associated words can appear, which may have a bigger impact. Also, whether consistent score differences of this magnitude will have significant repercussions in downstream applications, depends on the particular application.
Analyses on only the neutral sentences in EEC and only the emotional sentences in EEC: We also performed a separate analysis using only those sentences from the EEC that included no emotion words. Recall that there are four templates that contain no emotion words. Tables TABREF26 shows these results. We observe similar trends as in the analysis on the full set. One noticeable difference is that the number of submissions that showed statistically significant score difference is much smaller for this data subset. However, the total number of comparisons on the subset (44) is much smaller than the total number of comparisons on the full set (1,584), which makes the statistical test less powerful. Note also that the average score differences on the subset (columns 3 and 4 in Table TABREF26 ) tend to be higher than the differences on the full set (columns 3 and 4 in Table TABREF24 ). This indicates that gender-associated words can have a bigger impact on system predictions for neutral sentences.
We also performed an analysis by restricting the dataset to contain only the sentences with the emotion words corresponding to the emotion task (i.e., submissions to the anger intensity prediction task were evaluated only on sentences with anger words). The results (not shown here) were similar to the results on the full set.
Race Bias Results
We did a similar analysis for race as we did for gender. For each submission on each task, we calculated the difference between the average predicted score on the set of sentences with African American (AA) names and the average predicted score on the set of sentences with European American (EA) names. Then, we aggregated the results over all such sentence pairs in the EEC.
Table TABREF29 shows the results. The table has the same form and structure as the gender result tables. Observe that the number of submissions with no statistically significant score difference for sentences pertaining to the two races is about 5–11 (about 11% to 24%) for the four emotions and 3 (about 8%) for valence. These numbers are even lower than what was found for gender.
The majority of the systems assigned higher scores to sentences with African American names on the tasks of anger, fear, and sadness intensity prediction. On the joy and valence tasks, most submissions tended to assign higher scores to sentences with European American names. These tendencies reflect some common stereotypes that associate African Americans with more negative emotions BIBREF28 .
Figure FIGREF28 shows the score differences for individual systems on race sentence pairs on the valence regression task. Plots for the four emotion intensity prediction tasks are shown in Figure FIGREF32 in the Appendix. Here, the INLINEFORM0 –spreads are smaller than on the gender sentence pairs—from 0 to 0.15. As in the gender analysis, on the valence task the top 13 systems as well as some of the worst performing systems have smaller INLINEFORM1 –spread while the systems with medium to low performance show greater sensitivity to the race-associated names. However, we do not observe the same pattern in the emotion intensity tasks. Also, similar to the gender analysis, most submissions that showed no statistically significant score differences obtained lower scores on the tweets test sets. Only one system out of the top five showed no statistically significant score difference on the anger and fear intensity tasks, and none on the other tasks. Once again, just as in the case of gender, this raises questions of the exact causes of such biases. We hope to explore this in future work.
Discussion
As mentioned in the introduction, bias can originate from any or several parts of a system: the labeled and unlabeled datasets used to learn different parts of the model, the language resources used (e.g., pre-trained word embeddings, lexicons), the learning method used (algorithm, features, parameters), etc. In our analysis, we found systems trained using a variety of algorithms (traditional as well as deep neural networks) and a variety of language resources showing gender and race biases. Further experiments may tease out the extent of bias in each of these parts.
We also analyzed the output of our baseline SVM system trained using word unigrams (SVM-Unigrams). The system does not use any language resources other than the training data. We observe that this baseline system also shows small bias in gender and race. The INLINEFORM0 -spreads for this system were quite small: 0.09 to 0.2 on the gender sentence pairs and less than 0.002 on the race sentence pairs. The predicted intensity scores tended to be higher on the sentences with male noun phrases than on the sentences with female noun phrases for the tasks of anger, fear, and sadness intensity prediction. This tendency was reversed on the task of valence prediction. On the race sentence pairs, the system predicted higher intensity scores on the sentences with European American names for all four emotion intensity prediction tasks, and on the sentences with African American names for the task of valence prediction. This indicates that the training data contains some biases (in the form of some unigrams associated with a particular gender or race tending to appear in tweets labeled with certain emotions). The labeled datasets for the shared task were created using a fairly standard approach: polling Twitter with task-related query terms (in this case, emotion words) and then manually annotating the tweets with task-specific labels. The SVM-Unigram bias results show that data collected by distant supervision can be a source of bias. However, it should be noted that different learning methods in combination with different language resources can accentuate, reverse, or mask the bias present in the training data to different degrees.
Conclusions and Future Work
We created the Equity Evaluation Corpus (EEC), which consists of 8,640 sentences specifically chosen to tease out gender and race biases in natural language processing systems. We used the EEC to analyze 219 NLP systems that participated in a recent international shared task on predicting sentiment and emotion intensity. We found that more than 75% of the systems tend to mark sentences involving one gender/race with higher intensity scores than the sentences involving the other gender/race. We found such biases to be more widely prevalent for race than for gender. We also found that the bias can be different depending on the particular affect dimension involved.
We found the score differences across genders and across races to be somewhat small on average ( INLINEFORM0 , which is INLINEFORM1 of the 0 to 1 score range). However, for some systems the score differences reached as high as 0.34 (34%). What impact a consistent bias, even with an average magnitude INLINEFORM2 , might have in downstream applications merits further investigation.
We plan to extend the EEC with sentences associated with country names, professions (e.g., doctors, police officers, janitors, teachers, etc.), fields of study (e.g., arts vs. sciences), as well as races (e.g., Asian, mixed, etc.) and genders (e.g., agender, androgyne, trans, queer, etc.) not included in the current study. We can then use the corpus to examine biases across each of those variables as well. We are also interested in exploring which systems (or what techniques) accentuate inappropriate biases in the data and which systems mitigate such biases. Finally, we are interested in exploring how the quality of sentiment analysis predictions varies when applied to text produced by different demographic groups, such as people of different races, genders, and ethnicities.
The Equity Evaluation Corpus and the proposed methodology to examine bias are not meant to be comprehensive. However, using several approaches and datasets such as the one proposed here can bring about a more thorough examination of inappropriate biases in modern machine learning systems.
Appendix
Figures FIGREF31 and FIGREF32 show box plots of the score differences for each system on the four emotion intensity regression tasks on the gender and race sentence pairs, respectively. Each point on a plot corresponds to the difference in scores predicted by the system on one sentence pair. The systems are ordered by their performance rank (from first to last) on the task as per the official evaluation metric on the tweets test sets.
Question:
What criteria are used to select the 8,640 English sentences?
Answer:
| Sentences involving at least one race- or gender-associated word, sentence have to be short and grammatically simple, sentence have to include expressions of sentiment and emotion. | 328 | 30,036 | 30,038 | 30,109 | ... [The rest of the article is omitted]
|
0f12dc077fe8e5b95ca9163cea1dd17195c96929 | 0f12dc077fe8e5b95ca9163cea1dd17195c96929_1 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
[0]leftmargin=* [0]leftmargin=*
Automatic systems have had a significant and beneficial impact on all walks of human life. So much so that it is easy to overlook their potential to benefit society by promoting equity, diversity, and fairness. For example, machines do not take bribes to do their jobs, they can determine eligibility for a loan without being influenced by the color of the applicant's skin, and they can provide access to information and services without discrimination based on gender or sexual orientation. Nonetheless, as machine learning systems become more human-like in their predictions, they can also perpetuate human biases. Some learned biases may be beneficial for the downstream application (e.g., learning that humans often use some insect names, such as spider or cockroach, to refer to unpleasant situations). Other biases can be inappropriate and result in negative experiences for some groups of people. Examples include, loan eligibility and crime recidivism prediction systems that negatively assess people belonging to a certain pin/zip code (which may disproportionately impact people of a certain race) BIBREF0 and resumé sorting systems that believe that men are more qualified to be programmers than women BIBREF1 . Similarly, sentiment and emotion analysis systems can also perpetuate and accentuate inappropriate human biases, e.g., systems that consider utterances from one race or gender to be less positive simply because of their race or gender, or customer support systems that prioritize a call from an angry male over a call from the equally angry female.
Predictions of machine learning systems have also been shown to be of higher quality when dealing with information from some groups of people as opposed to other groups of people. For example, in the area of computer vision, gender classification systems perform particularly poorly for darker skinned females BIBREF2 . Natural language processing (NLP) systems have been shown to be poor in understanding text produced by people belonging to certain races BIBREF3 , BIBREF4 . For NLP systems, the sources of the bias often include the training data, other corpora, lexicons, and word embeddings that the machine learning algorithm may leverage to build its prediction model.
Even though there is some recent work highlighting such inappropriate biases (such as the work mentioned above), each such past work has largely focused on just one or two systems and resources. Further, there is no benchmark dataset for examining inappropriate biases in natural language systems. In this paper, we describe how we compiled a dataset of 8,640 English sentences carefully chosen to tease out biases towards certain races and genders. We will refer to it as the Equity Evaluation Corpus (EEC). We used the EEC as a supplementary test set in a recent shared task on predicting sentiment and emotion intensity in tweets, SemEval-2018 Task 1: Affect in Tweets BIBREF5 . In particular, we wanted to test a hypothesis that a system should equally rate the intensity of the emotion expressed by two sentences that differ only in the gender/race of a person mentioned. Note that here the term system refers to the combination of a machine learning architecture trained on a labeled dataset, and possibly using additional language resources. The bias can originate from any or several of these parts. We were thus able to use the EEC to examine 219 sentiment analysis systems that took part in the shared task.
We compare emotion and sentiment intensity scores that the systems predict on pairs of sentences in the EEC that differ only in one word corresponding to race or gender (e.g., `This man made me feel angry' vs. `This woman made me feel angry'). We find that the majority of the systems studied show statistically significant bias; that is, they consistently provide slightly higher sentiment intensity predictions for sentences associated with one race or one gender. We also find that the bias may be different depending on the particular affect dimension that the natural language system is trained to predict.
Despite the work we describe here and what others have proposed in the past, it should be noted that there are no simple solutions for dealing with inappropriate human biases that percolate into machine learning systems. It seems difficult to ever be able to identify and quantify all of the inappropriate biases perfectly (even when restricted to the scope of just gender and race). Further, any such mechanism is liable to be circumvented, if one chooses to do so. Nonetheless, as developers of sentiment analysis systems, and NLP systems more broadly, we cannot absolve ourselves of the ethical implications of the systems we build. Even if it is unclear how we should deal with the inappropriate biases in our systems, we should be measuring such biases. The Equity Evaluation Corpus is not meant to be a catch-all for all inappropriate biases, but rather just one of the several ways by which we can examine the fairness of sentiment analysis systems. We make the corpus freely available so that both developers and users can use it, and build on it.
Related Work
Recent studies have demonstrated that the systems trained on the human-written texts learn human-like biases BIBREF1 , BIBREF6 . In general, any predictive model built on historical data may inadvertently inherit human biases based on gender, ethnicity, race, or religion BIBREF7 , BIBREF8 . Discrimination-aware data mining focuses on measuring discrimination in data as well as on evaluating performance of discrimination-aware predictive models BIBREF9 , BIBREF10 , BIBREF11 , BIBREF12 .
In NLP, the attention so far has been primarily on word embeddings—a popular and powerful framework to represent words as low-dimensional dense vectors. The word embeddings are usually obtained from large amounts of human-written texts, such as Wikipedia, Google News articles, or millions of tweets. Bias in sentiment analysis systems has only been explored in simple systems that make use of pre-computed word embeddings BIBREF13 . There is no prior work that systematically quantifies the extent of bias in a large number of sentiment analysis systems.
This paper does not examine the differences in accuracies of systems on text produced by different races or genders, as was done by hovy2015demographic,blodgett2016demographic,jurgens2017incorporating,buolamwini2018gender. Approaches on how to mitigate inappropriate biases BIBREF14 , BIBREF1 , BIBREF15 , BIBREF16 , BIBREF13 , BIBREF17 , BIBREF18 are also beyond the scope of this paper. See also the position paper by hovy2016social, which identifies socio-ethical implications of the NLP systems in general.
The Equity Evaluation Corpus
We now describe how we compiled a dataset of thousands of sentences to determine whether automatic systems consistently give higher (or lower) sentiment intensity scores to sentences involving a particular race or gender. There are several ways in which such a dataset may be compiled. We present below the choices that we made.
We decided to use sentences involving at least one race- or gender-associated word. The sentences were intended to be short and grammatically simple. We also wanted some sentences to include expressions of sentiment and emotion, since the goal is to test sentiment and emotion systems. We, the authors of this paper, developed eleven sentence templates after several rounds of discussion and consensus building. They are shown in Table TABREF3 . The templates are divided into two groups. The first type (templates 1–7) includes emotion words. The purpose of this set is to have sentences expressing emotions. The second type (templates 8–11) does not include any emotion words. The purpose of this set is to have non-emotional (neutral) sentences.
The templates include two variables: INLINEFORM0 person INLINEFORM1 and INLINEFORM2 emotion word INLINEFORM3 . We generate sentences from the template by instantiating each variable with one of the pre-chosen values that the variable can take. Each of the eleven templates includes the variable INLINEFORM4 person INLINEFORM5 . INLINEFORM6 person INLINEFORM7 can be instantiated by any of the following noun phrases:
For our study, we chose ten names of each kind from the study by Caliskan:2017 (see Table TABREF7 ). The full lists of noun phrases representing females and males, used in our study, are shown in Table TABREF8 .
The second variable, INLINEFORM0 emotion word INLINEFORM1 , has two variants. Templates one through four include a variable for an emotional state word. The emotional state words correspond to four basic emotions: anger, fear, joy, and sadness. Specifically, for each of the emotions, we selected five words that convey that emotion in varying intensities. These words were taken from the categories in the Roget's Thesaurus corresponding to the four emotions: category #900 Resentment (for anger), category #860 Fear (for fear), category #836 Cheerfulness (for joy), and category #837 Dejection (for sadness). Templates five through seven include emotion words describing a situation or event. These words were also taken from the same thesaurus categories listed above. The full lists of emotion words (emotional state words and emotional situation/event words) are shown in Table TABREF10 .
We generated sentences from the templates by replacing INLINEFORM0 person INLINEFORM1 and INLINEFORM2 emotion word INLINEFORM3 variables with the values they can take. In total, 8,640 sentences were generated with the various combinations of INLINEFORM4 person INLINEFORM5 and INLINEFORM6 emotion word INLINEFORM7 values across the eleven templates. We manually examined the sentences to make sure they were grammatically well-formed. Notably, one can derive pairs of sentences from the EEC such that they differ only in one word corresponding to gender or race (e.g., `My daughter feels devastated' and `My son feels devastated'). We refer to the full set of 8,640 sentences as Equity Evaluation Corpus.
Measuring Race and Gender Bias in Automatic Sentiment Analysis Systems
The race and gender bias evaluation was carried out on the output of the 219 automatic systems that participated in SemEval-2018 Task 1: Affect in Tweets BIBREF5 . The shared task included five subtasks on inferring the affectual state of a person from their tweet: 1. emotion intensity regression, 2. emotion intensity ordinal classification, 3. valence (sentiment) regression, 4. valence ordinal classification, and 5. emotion classification. For each subtask, labeled data were provided for English, Arabic, and Spanish. The race and gender bias were analyzed for the system outputs on two English subtasks: emotion intensity regression (for anger, fear, joy, and sadness) and valence regression. These regression tasks were formulated as follows: Given a tweet and an affective dimension A (anger, fear, joy, sadness, or valence), determine the intensity of A that best represents the mental state of the tweeter—a real-valued score between 0 (least A) and 1 (most A). Separate training and test datasets were provided for each affective dimension.
Training sets included tweets along with gold intensity scores. Two test sets were provided for each task: 1. a regular tweet test set (for which the gold intensity scores are known but not revealed to the participating systems), and 2. the Equity Evaluation Corpus (for which no gold intensity labels exist). Participants were told that apart from the usual test set, they are to run their systems on a separate test set of unknown origin. The participants were instructed to train their system on the tweets training sets provided, and that they could use any other resources they may find or create. They were to run the same final system on the two test sets. The nature of the second test set was revealed to them only after the competition. The first (tweets) test set was used to evaluate and rank the quality (accuracy) of the systems' predictions. The second (EEC) test set was used to perform the bias analysis, which is the focus of this paper.
Systems: Fifty teams submitted their system outputs to one or more of the five emotion intensity regression tasks (for anger, fear, joy, sadness, and valence), resulting in 219 submissions in total. Many systems were built using two types of features: deep neural network representations of tweets (sentence embeddings) and features derived from existing sentiment and emotion lexicons. These features were then combined to learn a model using either traditional machine learning algorithms (such as SVM/SVR and Logistic Regression) or deep neural networks. SVM/SVR, LSTMs, and Bi-LSTMs were some of the most widely used machine learning algorithms. The sentence embeddings were obtained by training a neural network on the provided training data, a distant supervision corpus (e.g., AIT2018 Distant Supervision Corpus that has tweets with emotion-related query terms), sentiment-labeled tweet corpora (e.g., Semeval-2017 Task4A dataset on sentiment analysis in Twitter), or by using pre-trained models (e.g., DeepMoji BIBREF20 , Skip thoughts BIBREF21 ). The lexicon features were often derived from the NRC emotion and sentiment lexicons BIBREF22 , BIBREF23 , BIBREF24 , AFINN BIBREF25 , and Bing Liu Lexicon BIBREF26 .
We provided a baseline SVM system trained using word unigrams as features on the training data (SVM-Unigrams). This system is also included in the current analysis.
Measuring bias: To examine gender bias, we compared each system's predicted scores on the EEC sentence pairs as follows:
Thus, eleven pairs of scores (ten pairs of scores from ten noun phrase pairs and one pair of scores from the averages on name subsets) were examined for each template–emotion word instantiation. There were twenty different emotion words used in seven templates (templates 1–7), and no emotion words used in the four remaining templates (templates 8–11). In total, INLINEFORM0 pairs of scores were compared.
Similarly, to examine race bias, we compared pairs of system predicted scores as follows:
Thus, one pair of scores was examined for each template–emotion word instantiation. In total, INLINEFORM0 pairs of scores were compared.
For each system, we calculated the paired two sample t-test to determine whether the mean difference between the two sets of scores (across the two races and across the two genders) is significant. We set the significance level to 0.05. However, since we performed 438 assessments (219 submissions evaluated for biases in both gender and race), we applied Bonferroni correction. The null hypothesis that the true mean difference between the paired samples was zero was rejected if the calculated p-value fell below INLINEFORM0 .
Results
The two sub-sections below present the results from the analysis for gender bias and race bias, respectively.
Gender Bias Results
Individual submission results were communicated to the participants. Here, we present the summary results across all the teams. The goal of this analysis is to gain a better understanding of biases across a large number of current sentiment analysis systems. Thus, we partition the submissions into three groups according to the bias they show:
F=M not significant: submissions that showed no statistically significant difference in intensity scores predicted for corresponding female and male noun phrase sentences,
F INLINEFORM0 –M INLINEFORM1 significant: submissions that consistently gave higher scores for sentences with female noun phrases than for corresponding sentences with male noun phrases,
F INLINEFORM0 –M INLINEFORM1 significant: submissions that consistently gave lower scores for sentences with female noun phrases than for corresponding sentences with male noun phrases.
For each system and each sentence pair, we calculate the score difference INLINEFORM0 as the score for the female noun phrase sentence minus the score for the corresponding male noun phrase sentence. Table TABREF24 presents the summary results for each of the bias groups. It has the following columns:
#Subm.: number of submissions in each group.
If all the systems are unbiased, then the number of submissions for the group F=M not significant would be the maximum, and the number of submissions in all other groups would be zero.
Avg. score difference F INLINEFORM0 –M INLINEFORM1 : the average INLINEFORM2 for only those pairs where the score for the female noun phrase sentence is higher. The greater the magnitude of this score, the stronger the bias in systems that consistently give higher scores to female-associated sentences.
Avg. score difference F INLINEFORM0 –M INLINEFORM1 : the average INLINEFORM2 for only those pairs where the score for the female noun phrase sentence is lower. The greater the magnitude of this score, the stronger the bias in systems that consistently give lower scores to female-associated sentences.
Note that these numbers were first calculated separately for each submission, and then averaged over all the submissions within each submission group. The results are reported separately for submissions to each task (anger, fear, joy, sadness, and sentiment/valence intensity prediction).
Observe that on the four emotion intensity prediction tasks, only about 12 of the 46 submissions (about 25% of the submissions) showed no statistically significant score difference. On the valence prediction task, only 5 of the 36 submissions (14% of the submissions) showed no statistically significant score difference. Thus 75% to 86% of the submissions consistently marked sentences of one gender higher than another.
When predicting anger, joy, or valence, the number of systems consistently giving higher scores to sentences with female noun phrases (21–25) is markedly higher than the number of systems giving higher scores to sentences with male noun phrases (8–13). (Recall that higher valence means more positive sentiment.) In contrast, on the fear task, most submissions tended to assign higher scores to sentences with male noun phrases (23) as compared to the number of systems giving higher scores to sentences with female noun phrases (12). When predicting sadness, the number of submissions that mostly assigned higher scores to sentences with female noun phrases (18) is close to the number of submissions that mostly assigned higher scores to sentences with male noun phrases (16). These results are in line with some common stereotypes, such as females are more emotional, and situations involving male agents are more fearful BIBREF27 .
Figure FIGREF25 shows the score differences ( INLINEFORM0 ) for individual systems on the valence regression task. Plots for the four emotion intensity prediction tasks are shown in Figure FIGREF31 in the Appendix. Each point ( ▲, ▼, ●) on the plot corresponds to the difference in scores predicted by the system on one sentence pair. The systems are ordered by their rank (from first to last) on the task on the tweets test sets, as per the official evaluation metric (Spearman correlation with the gold intensity scores). We will refer to the difference between the maximal value of INLINEFORM1 and the minimal value of INLINEFORM2 for a particular system as the INLINEFORM3 –spread. Observe that the INLINEFORM4 –spreads for many systems are rather large, up to 0.57. Depending on the task, the top 10 or top 15 systems as well as some of the worst performing systems tend to have smaller INLINEFORM5 –spreads while the systems with medium to low performance show greater sensitivity to the gender-associated words. Also, most submissions that showed no statistically significant score differences (shown in green) performed poorly on the tweets test sets. Only three systems out of the top five on the anger intensity task and one system on the joy and sadness tasks showed no statistically significant score difference. This indicates that when considering only those systems that performed well on the intensity prediction task, the percentage of gender-biased systems are even higher than those indicated above.
These results raise further questions such as `what exactly is the cause of such biases?' and `why is the bias impacted by the emotion task under consideration?'. Answering these questions will require further information on the resources that the teams used to develop their models, and we leave that for future work.
Average score differences: For submissions that showed statistically significant score differences, the average score difference F INLINEFORM0 –M INLINEFORM1 and the average score difference F INLINEFORM2 –M INLINEFORM3 were INLINEFORM4 . Since the intensity scores range from 0 to 1, 0.03 is 3% of the full range. The maximal score difference ( INLINEFORM5 ) across all the submissions was as high as 0.34. Note, however, that these INLINEFORM6 s are the result of changing just one word in a sentence. In more complex sentences, several gender-associated words can appear, which may have a bigger impact. Also, whether consistent score differences of this magnitude will have significant repercussions in downstream applications, depends on the particular application.
Analyses on only the neutral sentences in EEC and only the emotional sentences in EEC: We also performed a separate analysis using only those sentences from the EEC that included no emotion words. Recall that there are four templates that contain no emotion words. Tables TABREF26 shows these results. We observe similar trends as in the analysis on the full set. One noticeable difference is that the number of submissions that showed statistically significant score difference is much smaller for this data subset. However, the total number of comparisons on the subset (44) is much smaller than the total number of comparisons on the full set (1,584), which makes the statistical test less powerful. Note also that the average score differences on the subset (columns 3 and 4 in Table TABREF26 ) tend to be higher than the differences on the full set (columns 3 and 4 in Table TABREF24 ). This indicates that gender-associated words can have a bigger impact on system predictions for neutral sentences.
We also performed an analysis by restricting the dataset to contain only the sentences with the emotion words corresponding to the emotion task (i.e., submissions to the anger intensity prediction task were evaluated only on sentences with anger words). The results (not shown here) were similar to the results on the full set.
Race Bias Results
We did a similar analysis for race as we did for gender. For each submission on each task, we calculated the difference between the average predicted score on the set of sentences with African American (AA) names and the average predicted score on the set of sentences with European American (EA) names. Then, we aggregated the results over all such sentence pairs in the EEC.
Table TABREF29 shows the results. The table has the same form and structure as the gender result tables. Observe that the number of submissions with no statistically significant score difference for sentences pertaining to the two races is about 5–11 (about 11% to 24%) for the four emotions and 3 (about 8%) for valence. These numbers are even lower than what was found for gender.
The majority of the systems assigned higher scores to sentences with African American names on the tasks of anger, fear, and sadness intensity prediction. On the joy and valence tasks, most submissions tended to assign higher scores to sentences with European American names. These tendencies reflect some common stereotypes that associate African Americans with more negative emotions BIBREF28 .
Figure FIGREF28 shows the score differences for individual systems on race sentence pairs on the valence regression task. Plots for the four emotion intensity prediction tasks are shown in Figure FIGREF32 in the Appendix. Here, the INLINEFORM0 –spreads are smaller than on the gender sentence pairs—from 0 to 0.15. As in the gender analysis, on the valence task the top 13 systems as well as some of the worst performing systems have smaller INLINEFORM1 –spread while the systems with medium to low performance show greater sensitivity to the race-associated names. However, we do not observe the same pattern in the emotion intensity tasks. Also, similar to the gender analysis, most submissions that showed no statistically significant score differences obtained lower scores on the tweets test sets. Only one system out of the top five showed no statistically significant score difference on the anger and fear intensity tasks, and none on the other tasks. Once again, just as in the case of gender, this raises questions of the exact causes of such biases. We hope to explore this in future work.
Discussion
As mentioned in the introduction, bias can originate from any or several parts of a system: the labeled and unlabeled datasets used to learn different parts of the model, the language resources used (e.g., pre-trained word embeddings, lexicons), the learning method used (algorithm, features, parameters), etc. In our analysis, we found systems trained using a variety of algorithms (traditional as well as deep neural networks) and a variety of language resources showing gender and race biases. Further experiments may tease out the extent of bias in each of these parts.
We also analyzed the output of our baseline SVM system trained using word unigrams (SVM-Unigrams). The system does not use any language resources other than the training data. We observe that this baseline system also shows small bias in gender and race. The INLINEFORM0 -spreads for this system were quite small: 0.09 to 0.2 on the gender sentence pairs and less than 0.002 on the race sentence pairs. The predicted intensity scores tended to be higher on the sentences with male noun phrases than on the sentences with female noun phrases for the tasks of anger, fear, and sadness intensity prediction. This tendency was reversed on the task of valence prediction. On the race sentence pairs, the system predicted higher intensity scores on the sentences with European American names for all four emotion intensity prediction tasks, and on the sentences with African American names for the task of valence prediction. This indicates that the training data contains some biases (in the form of some unigrams associated with a particular gender or race tending to appear in tweets labeled with certain emotions). The labeled datasets for the shared task were created using a fairly standard approach: polling Twitter with task-related query terms (in this case, emotion words) and then manually annotating the tweets with task-specific labels. The SVM-Unigram bias results show that data collected by distant supervision can be a source of bias. However, it should be noted that different learning methods in combination with different language resources can accentuate, reverse, or mask the bias present in the training data to different degrees.
Conclusions and Future Work
We created the Equity Evaluation Corpus (EEC), which consists of 8,640 sentences specifically chosen to tease out gender and race biases in natural language processing systems. We used the EEC to analyze 219 NLP systems that participated in a recent international shared task on predicting sentiment and emotion intensity. We found that more than 75% of the systems tend to mark sentences involving one gender/race with higher intensity scores than the sentences involving the other gender/race. We found such biases to be more widely prevalent for race than for gender. We also found that the bias can be different depending on the particular affect dimension involved.
We found the score differences across genders and across races to be somewhat small on average ( INLINEFORM0 , which is INLINEFORM1 of the 0 to 1 score range). However, for some systems the score differences reached as high as 0.34 (34%). What impact a consistent bias, even with an average magnitude INLINEFORM2 , might have in downstream applications merits further investigation.
We plan to extend the EEC with sentences associated with country names, professions (e.g., doctors, police officers, janitors, teachers, etc.), fields of study (e.g., arts vs. sciences), as well as races (e.g., Asian, mixed, etc.) and genders (e.g., agender, androgyne, trans, queer, etc.) not included in the current study. We can then use the corpus to examine biases across each of those variables as well. We are also interested in exploring which systems (or what techniques) accentuate inappropriate biases in the data and which systems mitigate such biases. Finally, we are interested in exploring how the quality of sentiment analysis predictions varies when applied to text produced by different demographic groups, such as people of different races, genders, and ethnicities.
The Equity Evaluation Corpus and the proposed methodology to examine bias are not meant to be comprehensive. However, using several approaches and datasets such as the one proposed here can bring about a more thorough examination of inappropriate biases in modern machine learning systems.
Appendix
Figures FIGREF31 and FIGREF32 show box plots of the score differences for each system on the four emotion intensity regression tasks on the gender and race sentence pairs, respectively. Each point on a plot corresponds to the difference in scores predicted by the system on one sentence pair. The systems are ordered by their performance rank (from first to last) on the task as per the official evaluation metric on the tweets test sets.
Question:
What criteria are used to select the 8,640 English sentences?
Answer:
| generated with the various combinations of INLINEFORM4 person INLINEFORM5 and INLINEFORM6 emotion word INLINEFORM7 values across the eleven templates, differ only in one word corresponding to gender or race | 328 | 30,036 | 30,038 | 30,109 | ... [The rest of the article is omitted]
|
518dae6f936882152c162058895db4eca815e649 | 518dae6f936882152c162058895db4eca815e649_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
This work is licenced under a Creative Commons Attribution 4.0 International License. License details: http://creativecommons.org/licenses/by/4.0/ Deep neural networks have been widely used in text classification and have achieved promising results BIBREF0 , BIBREF1 , BIBREF2 . Most focus on content information and use models such as convolutional neural networks (CNN) BIBREF3 or recursive neural networks BIBREF4 . However, for user-generated posts on social media like Facebook or Twitter, there is more information that should not be ignored. On social media platforms, a user can act either as the author of a post or as a reader who expresses his or her comments about the post.
In this paper, we classify posts taking into account post authorship, likes, topics, and comments. In particular, users and their “likes” hold strong potential for text mining. For example, given a set of posts that are related to a specific topic, a user's likes and dislikes provide clues for stance labeling. From a user point of view, users with positive attitudes toward the issue leave positive comments on the posts with praise or even just the post's content; from a post point of view, positive posts attract users who hold positive stances. We also investigate the influence of topics: different topics are associated with different stance labeling tendencies and word usage. For example we discuss women's rights and unwanted babies on the topic of abortion, but we criticize medicine usage or crime when on the topic of marijuana BIBREF5 . Even for posts on a specific topic like nuclear power, a variety of arguments are raised: green energy, radiation, air pollution, and so on. As for comments, we treat them as additional text information. The arguments in the comments and the commenters (the users who leave the comments) provide hints on the post's content and further facilitate stance classification.
In this paper, we propose the user-topic-comment neural network (UTCNN), a deep learning model that utilizes user, topic, and comment information. We attempt to learn user and topic representations which encode user interactions and topic influences to further enhance text classification, and we also incorporate comment information. We evaluate this model on a post stance classification task on forum-style social media platforms. The contributions of this paper are as follows: 1. We propose UTCNN, a neural network for text in modern social media channels as well as legacy social media, forums, and message boards — anywhere that reveals users, their tastes, as well as their replies to posts. 2. When classifying social media post stances, we leverage users, including authors and likers. User embeddings can be generated even for users who have never posted anything. 3. We incorporate a topic model to automatically assign topics to each post in a single topic dataset. 4. We show that overall, the proposed method achieves the highest performance in all instances, and that all of the information extracted, whether users, topics, or comments, still has its contributions.
Extra-Linguistic Features for Stance Classification
In this paper we aim to use text as well as other features to see how they complement each other in a deep learning model. In the stance classification domain, previous work has showed that text features are limited, suggesting that adding extra-linguistic constraints could improve performance BIBREF6 , BIBREF7 , BIBREF8 . For example, Hasan and Ng as well as Thomas et al. require that posts written by the same author have the same stance BIBREF9 , BIBREF10 . The addition of this constraint yields accuracy improvements of 1–7% for some models and datasets. Hasan and Ng later added user-interaction constraints and ideology constraints BIBREF7 : the former models the relationship among posts in a sequence of replies and the latter models inter-topic relationships, e.g., users who oppose abortion could be conservative and thus are likely to oppose gay rights.
For work focusing on online forum text, since posts are linked through user replies, sequential labeling methods have been used to model relationships between posts. For example, Hasan and Ng use hidden Markov models (HMMs) to model dependent relationships to the preceding post BIBREF9 ; Burfoot et al. use iterative classification to repeatedly generate new estimates based on the current state of knowledge BIBREF11 ; Sridhar et al. use probabilistic soft logic (PSL) to model reply links via collaborative filtering BIBREF12 . In the Facebook dataset we study, we use comments instead of reply links. However, as the ultimate goal in this paper is predicting not comment stance but post stance, we treat comments as extra information for use in predicting post stance.
Deep Learning on Extra-Linguistic Features
In recent years neural network models have been applied to document sentiment classification BIBREF13 , BIBREF4 , BIBREF14 , BIBREF15 , BIBREF2 . Text features can be used in deep networks to capture text semantics or sentiment. For example, Dong et al. use an adaptive layer in a recursive neural network for target-dependent Twitter sentiment analysis, where targets are topics such as windows 7 or taylor swift BIBREF16 , BIBREF17 ; recursive neural tensor networks (RNTNs) utilize sentence parse trees to capture sentence-level sentiment for movie reviews BIBREF4 ; Le and Mikolov predict sentiment by using paragraph vectors to model each paragraph as a continuous representation BIBREF18 . They show that performance can thus be improved by more delicate text models.
Others have suggested using extra-linguistic features to improve the deep learning model. The user-word composition vector model (UWCVM) BIBREF19 is inspired by the possibility that the strength of sentiment words is user-specific; to capture this they add user embeddings in their model. In UPNN, a later extension, they further add a product-word composition as product embeddings, arguing that products can also show different tendencies of being rated or reviewed BIBREF20 . Their addition of user information yielded 2–10% improvements in accuracy as compared to the above-mentioned RNTN and paragraph vector methods. We also seek to inject user information into the neural network model. In comparison to the research of Tang et al. on sentiment classification for product reviews, the difference is two-fold. First, we take into account multiple users (one author and potentially many likers) for one post, whereas only one user (the reviewer) is involved in a review. Second, we add comment information to provide more features for post stance classification. None of these two factors have been considered previously in a deep learning model for text stance classification. Therefore, we propose UTCNN, which generates and utilizes user embeddings for all users — even for those who have not authored any posts — and incorporates comments to further improve performance.
Method
In this section, we first describe CNN-based document composition, which captures user- and topic-dependent document-level semantic representation from word representations. Then we show how to add comment information to construct the user-topic-comment neural network (UTCNN).
User- and Topic-dependent Document Composition
As shown in Figure FIGREF4 , we use a general CNN BIBREF3 and two semantic transformations for document composition . We are given a document with an engaged user INLINEFORM0 , a topic INLINEFORM1 , and its composite INLINEFORM2 words, each word INLINEFORM3 of which is associated with a word embedding INLINEFORM4 where INLINEFORM5 is the vector dimension. For each word embedding INLINEFORM6 , we apply two dot operations as shown in Equation EQREF6 : DISPLAYFORM0
where INLINEFORM0 models the user reading preference for certain semantics, and INLINEFORM1 models the topic semantics; INLINEFORM2 and INLINEFORM3 are the dimensions of transformed user and topic embeddings respectively. We use INLINEFORM4 to model semantically what each user prefers to read and/or write, and use INLINEFORM5 to model the semantics of each topic. The dot operation of INLINEFORM6 and INLINEFORM7 transforms the global representation INLINEFORM8 to a user-dependent representation. Likewise, the dot operation of INLINEFORM9 and INLINEFORM10 transforms INLINEFORM11 to a topic-dependent representation.
After the two dot operations on INLINEFORM0 , we have user-dependent and topic-dependent word vectors INLINEFORM1 and INLINEFORM2 , which are concatenated to form a user- and topic-dependent word vector INLINEFORM3 . Then the transformed word embeddings INLINEFORM4 are used as the CNN input. Here we apply three convolutional layers on the concatenated transformed word embeddings INLINEFORM5 : DISPLAYFORM0
where INLINEFORM0 is the index of words; INLINEFORM1 is a non-linear activation function (we use INLINEFORM2 ); INLINEFORM5 is the convolutional filter with input length INLINEFORM6 and output length INLINEFORM7 , where INLINEFORM8 is the window size of the convolutional operation; and INLINEFORM9 and INLINEFORM10 are the output and bias of the convolution layer INLINEFORM11 , respectively. In our experiments, the three window sizes INLINEFORM12 in the three convolution layers are one, two, and three, encoding unigram, bigram, and trigram semantics accordingly.
After the convolutional layer, we add a maximum pooling layer among convolutional outputs to obtain the unigram, bigram, and trigram n-gram representations. This is succeeded by an average pooling layer for an element-wise average of the three maximized convolution outputs.
UTCNN Model Description
Figure FIGREF10 illustrates the UTCNN model. As more than one user may interact with a given post, we first add a maximum pooling layer after the user matrix embedding layer and user vector embedding layer to form a moderator matrix embedding INLINEFORM0 and a moderator vector embedding INLINEFORM1 for moderator INLINEFORM2 respectively, where INLINEFORM3 is used for the semantic transformation in the document composition process, as mentioned in the previous section. The term moderator here is to denote the pseudo user who provides the overall semantic/sentiment of all the engaged users for one document. The embedding INLINEFORM4 models the moderator stance preference, that is, the pattern of the revealed user stance: whether a user is willing to show his preference, whether a user likes to show impartiality with neutral statements and reasonable arguments, or just wants to show strong support for one stance. Ideally, the latent user stance is modeled by INLINEFORM5 for each user. Likewise, for topic information, a maximum pooling layer is added after the topic matrix embedding layer and topic vector embedding layer to form a joint topic matrix embedding INLINEFORM6 and a joint topic vector embedding INLINEFORM7 for topic INLINEFORM8 respectively, where INLINEFORM9 models the semantic transformation of topic INLINEFORM10 as in users and INLINEFORM11 models the topic stance tendency. The latent topic stance is also modeled by INLINEFORM12 for each topic.
As for comments, we view them as short documents with authors only but without likers nor their own comments. Therefore we apply document composition on comments although here users are commenters (users who comment). It is noticed that the word embeddings INLINEFORM0 for the same word in the posts and comments are the same, but after being transformed to INLINEFORM1 in the document composition process shown in Figure FIGREF4 , they might become different because of their different engaged users. The output comment representation together with the commenter vector embedding INLINEFORM2 and topic vector embedding INLINEFORM3 are concatenated and a maximum pooling layer is added to select the most important feature for comments. Instead of requiring that the comment stance agree with the post, UTCNN simply extracts the most important features of the comment contents; they could be helpful, whether they show obvious agreement or disagreement. Therefore when combining comment information here, the maximum pooling layer is more appropriate than other pooling or merging layers. Indeed, we believe this is one reason for UTCNN's performance gains.
Finally, the pooled comment representation, together with user vector embedding INLINEFORM0 , topic vector embedding INLINEFORM1 , and document representation are fed to a fully connected network, and softmax is applied to yield the final stance label prediction for the post.
Experiment
We start with the experimental dataset and then describe the training process as well as the implementation of the baselines. We also implement several variations to reveal the effects of features: authors, likers, comment, and commenters. In the results section we compare our model with related work.
Dataset
We tested the proposed UTCNN on two different datasets: FBFans and CreateDebate. FBFans is a privately-owned, single-topic, Chinese, unbalanced, social media dataset, and CreateDebate is a public, multiple-topic, English, balanced, forum dataset. Results using these two datasets show the applicability and superiority for different topics, languages, data distributions, and platforms.
The FBFans dataset contains data from anti-nuclear-power Chinese Facebook fan groups from September 2013 to August 2014, including posts and their author and liker IDs. There are a total of 2,496 authors, 505,137 likers, 33,686 commenters, and 505,412 unique users. Two annotators were asked to take into account only the post content to label the stance of the posts in the whole dataset as supportive, neutral, or unsupportive (hereafter denoted as Sup, Neu, and Uns). Sup/Uns posts were those in support of or against anti-reconstruction; Neu posts were those evincing a neutral standpoint on the topic, or were irrelevant. Raw agreement between annotators is 0.91, indicating high agreement. Specifically, Cohen’s Kappa for Neu and not Neu labeling is 0.58 (moderate), and for Sup or Uns labeling is 0.84 (almost perfect). Posts with inconsistent labels were filtered out, and the development and testing sets were randomly selected from what was left. Posts in the development and testing sets involved at least one user who appeared in the training set. The number of posts for each stance is shown on the left-hand side of Table TABREF12 . About twenty percent of the posts were labeled with a stance, and the number of supportive (Sup) posts was much larger than that of the unsupportive (Uns) ones: this is thus highly skewed data, which complicates stance classification. On average, 161.1 users were involved in one post. The maximum was 23,297 and the minimum was one (the author). For comments, on average there were 3 comments per post. The maximum was 1,092 and the minimum was zero.
To test whether the assumption of this paper – posts attract users who hold the same stance to like them – is reliable, we examine the likes from authors of different stances. Posts in FBFans dataset are used for this analysis. We calculate the like statistics of each distinct author from these 32,595 posts. As the numbers of authors in the Sup, Neu and Uns stances are largely imbalanced, these numbers are normalized by the number of users of each stance. Table TABREF13 shows the results. Posts with stances (i.e., not neutral) attract users of the same stance. Neutral posts also attract both supportive and neutral users, like what we observe in supportive posts, but just the neutral posts can attract even more neutral likers. These results do suggest that users prefer posts of the same stance, or at least posts of no obvious stance which might cause annoyance when reading, and hence support the user modeling in our approach.
The CreateDebate dataset was collected from an English online debate forum discussing four topics: abortion (ABO), gay rights (GAY), Obama (OBA), and marijuana (MAR). The posts are annotated as for (F) and against (A). Replies to posts in this dataset are also labeled with stance and hence use the same data format as posts. The labeling results are shown in the right-hand side of Table TABREF12 . We observe that the dataset is more balanced than the FBFans dataset. In addition, there are 977 unique users in the dataset. To compare with Hasan and Ng's work, we conducted five-fold cross-validation and present the annotation results as the average number of all folds BIBREF9 , BIBREF5 .
The FBFans dataset has more integrated functions than the CreateDebate dataset; thus our model can utilize all linguistic and extra-linguistic features. For the CreateDebate dataset, on the other hand, the like and comment features are not available (as there is a stance label for each reply, replies are evaluated as posts as other previous work) but we still implemented our model using the content, author, and topic information.
Settings
In the UTCNN training process, cross-entropy was used as the loss function and AdaGrad as the optimizer. For FBFans dataset, we learned the 50-dimensional word embeddings on the whole dataset using GloVe BIBREF21 to capture the word semantics; for CreateDebate dataset we used the publicly available English 50-dimensional word embeddings, pre-trained also using GloVe. These word embeddings were fixed in the training process. The learning rate was set to 0.03. All user and topic embeddings were randomly initialized in the range of [-0.1 0.1]. Matrix embeddings for users and topics were sized at 250 ( INLINEFORM0 ); vector embeddings for users and topics were set to length 10.
We applied the LDA topic model BIBREF22 on the FBFans dataset to determine the latent topics with which to build topic embeddings, as there is only one general known topic: nuclear power plants. We learned 100 latent topics and assigned the top three topics for each post. For the CreateDebate dataset, which itself constitutes four topics, the topic labels for posts were used directly without additionally applying LDA.
For the FBFans data we report class-based f-scores as well as the macro-average f-score ( INLINEFORM0 ) shown in equation EQREF19 . DISPLAYFORM0
where INLINEFORM0 and INLINEFORM1 are the average precision and recall of the three class. We adopted the macro-average f-score as the evaluation metric for the overall performance because (1) the experimental dataset is severely imbalanced, which is common for contentious issues; and (2) for stance classification, content in minor-class posts is usually more important for further applications. For the CreateDebate dataset, accuracy was adopted as the evaluation metric to compare the results with related work BIBREF7 , BIBREF9 , BIBREF12 .
Baselines
We pit our model against the following baselines: 1) SVM with unigram, bigram, and trigram features, which is a standard yet rather strong classifier for text features; 2) SVM with average word embedding, where a document is represented as a continuous representation by averaging the embeddings of the composite words; 3) SVM with average transformed word embeddings (the INLINEFORM0 in equation EQREF6 ), where a document is represented as a continuous representation by averaging the transformed embeddings of the composite words; 4) two mature deep learning models on text classification, CNN BIBREF3 and Recurrent Convolutional Neural Networks (RCNN) BIBREF0 , where the hyperparameters are based on their work; 5) the above SVM and deep learning models with comment information; 6) UTCNN without user information, representing a pure-text CNN model where we use the same user matrix and user embeddings INLINEFORM1 and INLINEFORM2 for each user; 7) UTCNN without the LDA model, representing how UTCNN works with a single-topic dataset; 8) UTCNN without comments, in which the model predicts the stance label given only user and topic information. All these models were trained on the training set, and parameters as well as the SVM kernel selections (linear or RBF) were fine-tuned on the development set. Also, we adopt oversampling on SVMs, CNN and RCNN because the FBFans dataset is highly imbalanced.
Results on FBFans Dataset
In Table TABREF22 we show the results of UTCNN and the baselines on the FBFans dataset. Here Majority yields good performance on Neu since FBFans is highly biased to the neutral class. The SVM models perform well on Sup and Neu but perform poorly for Uns, showing that content information in itself is insufficient to predict stance labels, especially for the minor class. With the transformed word embedding feature, SVM can achieve comparable performance as SVM with n-gram feature. However, the much fewer feature dimension of the transformed word embedding makes SVM with word embeddings a more efficient choice for modeling the large scale social media dataset. For the CNN and RCNN models, they perform slightly better than most of the SVM models but still, the content information is insufficient to achieve a good performance on the Uns posts. As to adding comment information to these models, since the commenters do not always hold the same stance as the author, simply adding comments and post contents together merely adds noise to the model.
Among all UTCNN variations, we find that user information is most important, followed by topic and comment information. UTCNN without user information shows results similar to SVMs — it does well for Sup and Neu but detects no Uns. Its best f-scores on both Sup and Neu among all methods show that with enough training data, content-based models can perform well; at the same time, the lack of user information results in too few clues for minor-class posts to either predict their stance directly or link them to other users and posts for improved performance. The 17.5% improvement when adding user information suggests that user information is especially useful when the dataset is highly imbalanced. All models that consider user information predict the minority class successfully. UCTNN without topic information works well but achieves lower performance than the full UTCNN model. The 4.9% performance gain brought by LDA shows that although it is satisfactory for single topic datasets, adding that latent topics still benefits performance: even when we are discussing the same topic, we use different arguments and supporting evidence. Lastly, we get 4.8% improvement when adding comment information and it achieves comparable performance to UTCNN without topic information, which shows that comments also benefit performance. For platforms where user IDs are pixelated or otherwise hidden, adding comments to a text model still improves performance. In its integration of user, content, and comment information, the full UTCNN produces the highest f-scores on all Sup, Neu, and Uns stances among models that predict the Uns class, and the highest macro-average f-score overall. This shows its ability to balance a biased dataset and supports our claim that UTCNN successfully bridges content and user, topic, and comment information for stance classification on social media text. Another merit of UTCNN is that it does not require a balanced training data. This is supported by its outperforming other models though no oversampling technique is applied to the UTCNN related experiments as shown in this paper. Thus we can conclude that the user information provides strong clues and it is still rich even in the minority class.
We also investigate the semantic difference when a user acts as an author/liker or a commenter. We evaluated a variation in which all embeddings from the same user were forced to be identical (this is the UTCNN shared user embedding setting in Table TABREF22 ). This setting yielded only a 2.5% improvement over the model without comments, which is not statistically significant. However, when separating authors/likers and commenters embeddings (i.e., the UTCNN full model), we achieved much greater improvements (4.8%). We attribute this result to the tendency of users to use different wording for different roles (for instance author vs commenter). This is observed when the user, acting as an author, attempts to support her argument against nuclear power by using improvements in solar power; when acting as a commenter, though, she interacts with post contents by criticizing past politicians who supported nuclear power or by arguing that the proposed evacuation plan in case of a nuclear accident is ridiculous. Based on this finding, in the final UTCNN setting we train two user matrix embeddings for one user: one for the author/liker role and the other for the commenter role.
Results on CreateDebate Dataset
Table TABREF24 shows the results of UTCNN, baselines as we implemented on the FBFans datset and related work on the CreateDebate dataset. We do not adopt oversampling on these models because the CreateDebate dataset is almost balanced. In previous work, integer linear programming (ILP) or linear-chain conditional random fields (CRFs) were proposed to integrate text features, author, ideology, and user-interaction constraints, where text features are unigram, bigram, and POS-dependencies; the author constraint tends to require that posts from the same author for the same topic hold the same stance; the ideology constraint aims to capture inferences between topics for the same author; the user-interaction constraint models relationships among posts via user interactions such as replies BIBREF7 , BIBREF9 .
The SVM with n-gram or average word embedding feature performs just similar to the majority. However, with the transformed word embedding, it achieves superior results. It shows that the learned user and topic embeddings really capture the user and topic semantics. This finding is not so obvious in the FBFans dataset and it might be due to the unfavorable data skewness for SVM. As for CNN and RCNN, they perform slightly better than most SVMs as we found in Table TABREF22 for FBFans.
Compared to the ILP BIBREF7 and CRF BIBREF9 methods, the UTCNN user embeddings encode author and user-interaction constraints, where the ideology constraint is modeled by the topic embeddings and text features are modeled by the CNN. The significant improvement achieved by UTCNN suggests the latent representations are more effective than overt model constraints.
The PSL model BIBREF12 jointly labels both author and post stance using probabilistic soft logic (PSL) BIBREF23 by considering text features and reply links between authors and posts as in Hasan and Ng's work. Table TABREF24 reports the result of their best AD setting, which represents the full joint stance/disagreement collective model on posts and is hence more relevant to UTCNN. In contrast to their model, the UTCNN user embeddings represent relationships between authors, but UTCNN models do not utilize link information between posts. Though the PSL model has the advantage of being able to jointly label the stances of authors and posts, its performance on posts is lower than the that for the ILP or CRF models. UTCNN significantly outperforms these models on posts and has the potential to predict user stances through the generated user embeddings.
For the CreateDebate dataset, we also evaluated performance when not using topic embeddings or user embeddings; as replies in this dataset are viewed as posts, the setting without comment embeddings is not available. Table TABREF24 shows the same findings as Table TABREF22 : the 21% improvement in accuracy demonstrates that user information is the most vital. This finding also supports the results in the related work: user constraints are useful and can yield 11.2% improvement in accuracy BIBREF7 . Further considering topic information yields 3.4% improvement, suggesting that knowing the subject of debates provides useful information. In sum, Table TABREF22 together with Table TABREF24 show that UTCNN achieves promising performance regardless of topic, language, data distribution, and platform.
Conclusion
We have proposed UTCNN, a neural network model that incorporates user, topic, content and comment information for stance classification on social media texts. UTCNN learns user embeddings for all users with minimum active degree, i.e., one post or one like. Topic information obtained from the topic model or the pre-defined labels further improves the UTCNN model. In addition, comment information provides additional clues for stance classification. We have shown that UTCNN achieves promising and balanced results. In the future we plan to explore the effectiveness of the UTCNN user embeddings for author stance classification.
Acknowledgements
Research of this paper was partially supported by Ministry of Science and Technology, Taiwan, under the contract MOST 104-2221-E-001-024-MY2.
Question:
How many layers does the UTCNN model have?
Answer:
| eight layers | 328 | 29,212 | 29,214 | 29,266 | ... [The rest of the article is omitted]
|
58ef2442450c392bfc55c4dc35f216542f5f2dbb | 58ef2442450c392bfc55c4dc35f216542f5f2dbb_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
The rapid growth in speech and small screen interfaces, particularly on mobile devices, has significantly influenced the way users interact with intelligent systems to satisfy their information needs. The growing interest in personal digital assistants, such as Amazon Alexa, Apple Siri, Google Assistant, and Microsoft Cortana, demonstrates the willingness of users to employ conversational interactions BIBREF0. As a result, conversational information seeking (CIS) has been recognized as a major emerging research area in the Third Strategic Workshop on Information Retrieval (SWIRL 2018) BIBREF1.
Research progress in CIS relies on the availability of resources to the community. There have been recent efforts on providing data for various CIS tasks, such as the TREC 2019 Conversational Assistance Track (CAsT), MISC BIBREF2, Qulac BIBREF3, CoQA BIBREF4, QuAC BIBREF5, SCS BIBREF6, and CCPE-M BIBREF7. In addition, BIBREF8 have implemented a demonstration for conversational movie recommendation based on Google's DialogFlow. Despite all of these resources, the community still feels the lack of a suitable platform for developing CIS systems. We believe that providing such platform will speed up the progress in conversational information seeking research. Therefore, we developed a general framework for supporting CIS research. The framework is called Macaw. This paper describes the high-level architecture of Macaw, the supported functionality, and our future vision. Researchers working on various CIS tasks should be able to take advantage of Macaw in their projects.
Macaw is designed based on a modular architecture to support different information seeking tasks, including conversational search, conversational question answering, conversational recommendation, and conversational natural language interface to structured and semi-structured data. Each interaction in Macaw (from both user and system) is a Message object, thus a conversation is a list of Messages. Macaw consists of multiple actions, each action is a module that can satisfy the information needs of users for some requests. For example, search and question answering can be two actions in Macaw. Even multiple search algorithms can be also seen as multiple actions. Each action can produce multiple outputs (e.g., multiple retrieved documents). For every user interaction, Macaw runs all actions in parallel. The actions' outputs produced within a predefined time interval (i.e., an interaction timeout constant) are then post-processed. Macaw can choose one or combine multiple of these outputs and prepare an output Message object as the user's response.
The modular design of Macaw makes it relatively easy to configure a different user interface or add a new one. The current implementation of Macaw supports a command line interface as well as mobile, desktop, and web apps. In more detail, Macaw's interface can be a Telegram bot, which supports a wide range of devices and operating systems (see FIGREF4). This allows Macaw to support multi-modal interactions, such as text, speech, image, click, etc. A number of APIs for automatic speech recognition and generation have been employed to support speech interactions. Note that the Macaw's architecture and implementation allows mixed-initiative interactions.
The research community can benefit from Macaw for the following purposes:
[leftmargin=*]
Developing algorithms, tools, and techniques for CIS.
Studying user interactions with CIS systems.
Performing CIS studies based on an intermediary person and wizard of oz.
Preparing quick demonstration for a developed CIS model.
Macaw Architecture
Macaw has a modular design, with the goal of making it easy to configure and add new modules such as a different user interface or different retrieval module. The overall setup also follows a Model-View-Controller (MVC) like architecture. The design decisions have been made to smooth the Macaw's adoptions and extensions. Macaw is implemented in Python, thus machine learning models implemented using PyTorch, Scikit-learn, or TensorFlow can be easily integrated into Macaw. The high-level overview of Macaw is depicted in FIGREF8. The user interacts with the interface and the interface produces a Message object from the current interaction of user. The interaction can be in multi-modal form, such as text, speech, image, and click. Macaw stores all interactions in an “Interaction Database”. For every interaction, Macaw looks for most recent user-system interactions (including the system's responses) to create a list of Messages, called the conversation list. It is then dispatched to multiple information seeking (and related) actions. The actions run in parallel, and each should respond within a pre-defined time interval. The output selection component selects from (or potentially combines) the outputs generated by different actions and creates a Message object as the system's response. This message is logged into the interaction database and is sent to the interface to be presented to the user. Again, the response message can be multi-modal and include text, speech, link, list of options, etc.
Macaw also supports Wizard of Oz studies or intermediary-based information seeking studies. The architecture of Macaw for such setup is presented in FIGREF16. As shown in the figure, the seeker interacts with a real conversational interface that supports multi-modal and mixed-initiative interactions in multiple devices. The intermediary (or the wizard) receives the seeker's message and performs different information seeking actions with Macaw. All seeker-intermediary and intermediary-system interactions will be logged for further analysis. This setup can simulate an ideal CIS system and thus is useful for collecting high-quality data from real users for CIS research.
Retrieval and Question Answering in Macaw
The overview of retrieval and question answering actions in Macaw is shown in FIGREF17. These actions consist of the following components:
[leftmargin=*]
Co-Reference Resolution: To support multi-turn interactions, it is sometimes necessary to use co-reference resolution techniques for effective retrieval. In Macaw, we identify all the co-references from the last request of user to the conversation history. The same co-reference resolution outputs can be used for different query generation components. This can be a generic or action-specific component.
Query Generation: This component generates a query based on the past user-system interactions. The query generation component may take advantage of co-reference resolution for query expansion or re-writing.
Retrieval Model: This is the core ranking component that retrieves documents or passages from a large collection. Macaw can retrieve documents from an arbitrary document collection using the Indri python interface BIBREF9, BIBREF10. We also provide the support for web search using the Bing Web Search API. Macaw also allows multi-stage document re-ranking.
Result Generation: The retrieved documents can be too long to be presented using some interfaces. Result generation is basically a post-processing step ran on the retrieved result list. In case of question answering, it can employ answer selection or generation techniques, such as machine reading comprehension models. For example, Macaw features the DrQA model BIBREF11 for question answering.
These components are implemented in a generic form, so researchers can easily replace them with their own favorite algorithms.
User Interfaces
We have implemented the following interfaces for Macaw:
[leftmargin=*]
File IO: This interface is designed for experimental purposes, such as evaluating the performance of a conversational search technique on a dataset with multiple queries. This is not an interactive interface.
Standard IO: This interactive command line interface is designed for development purposes to interact with the system, see the logs, and debug or improve the system.
Telegram: This interactive interface is designed for interaction with real users (see FIGREF4). Telegram is a popular instant messaging service whose client-side code is open-source. We have implemented a Telegram bot that can be used with different devices (personal computers, tablets, and mobile phones) and different operating systems (Android, iOS, Linux, Mac OS, and Windows). This interface allows multi-modal interactions (text, speech, click, image). It can be also used for speech-only interactions. For speech recognition and generation, Macaw relies on online APIs, e.g., the services provided by Google Cloud and Microsoft Azure. In addition, there exist multiple popular groups and channels in Telegram, which allows further integration of social networks with conversational systems. For example, see the Naseri and Zamani's study on news popularity in Telegram BIBREF12.
Similar to the other modules, one can easily extend Macaw using other appropriate user interfaces.
Limitations and Future Work
The current implementation of Macaw lacks the following actions. We intend to incrementally improve Macaw by supporting more actions and even more advanced techniques for the developed actions.
[leftmargin=*]
Clarification and Preference Elicitation: Asking clarifying questions has been recently recognized as a necessary component in a conversational system BIBREF3, BIBREF7. The authors are not aware of a published solution for generating clarifying questions using public resources. Therefore, Macaw does not currently support clarification.
Explanation: Despite its importance, result list explanation is also a relatively less explored topic. We intend to extend Macaw with result list explanation as soon as we find a stable and mature solution.
Recommendation: In our first release, we focus on conversational search and question answering tasks. We intend to provide support for conversational recommendation, e.g., BIBREF13, BIBREF14, BIBREF15, and joint search and recommendation, e.g., BIBREF16, BIBREF17, in the future.
Natural Language Interface: Macaw can potentially support access to structured data, such as knowledge graph. We would like to ease conversational natural language interface to structured and semi-structured data in our future releases.
Contribution
Macaw is distributed under the MIT License. We welcome contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com. This project has adopted the Microsoft Open Source Code of Conduct.
When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.
Conclusions
This paper described Macaw, an open-source platform for conversational information seeking research. Macaw supports multi-turn, multi-modal, and mixed-initiative interactions. It was designed based on a modular architecture that allows further improvements and extensions. Researchers can benefit from Macaw for developing algorithms and techniques for conversational information seeking research, for user studies with different interfaces, for data collection from real users, and for preparing a demonstration of a CIS model.
Acknowledgements
The authors wish to thank Ahmed Hassan Awadallah, Krisztian Balog, and Arjen P. de Vries for their invaluable feedback.
Question:
Does the paper provide any case studies to illustrate how one can use Macaw for CIS research?
Answer:
| No | 328 | 11,979 | 11,981 | 12,084 | ... [The rest of the article is omitted]
|
58ef2442450c392bfc55c4dc35f216542f5f2dbb | 58ef2442450c392bfc55c4dc35f216542f5f2dbb_1 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
The rapid growth in speech and small screen interfaces, particularly on mobile devices, has significantly influenced the way users interact with intelligent systems to satisfy their information needs. The growing interest in personal digital assistants, such as Amazon Alexa, Apple Siri, Google Assistant, and Microsoft Cortana, demonstrates the willingness of users to employ conversational interactions BIBREF0. As a result, conversational information seeking (CIS) has been recognized as a major emerging research area in the Third Strategic Workshop on Information Retrieval (SWIRL 2018) BIBREF1.
Research progress in CIS relies on the availability of resources to the community. There have been recent efforts on providing data for various CIS tasks, such as the TREC 2019 Conversational Assistance Track (CAsT), MISC BIBREF2, Qulac BIBREF3, CoQA BIBREF4, QuAC BIBREF5, SCS BIBREF6, and CCPE-M BIBREF7. In addition, BIBREF8 have implemented a demonstration for conversational movie recommendation based on Google's DialogFlow. Despite all of these resources, the community still feels the lack of a suitable platform for developing CIS systems. We believe that providing such platform will speed up the progress in conversational information seeking research. Therefore, we developed a general framework for supporting CIS research. The framework is called Macaw. This paper describes the high-level architecture of Macaw, the supported functionality, and our future vision. Researchers working on various CIS tasks should be able to take advantage of Macaw in their projects.
Macaw is designed based on a modular architecture to support different information seeking tasks, including conversational search, conversational question answering, conversational recommendation, and conversational natural language interface to structured and semi-structured data. Each interaction in Macaw (from both user and system) is a Message object, thus a conversation is a list of Messages. Macaw consists of multiple actions, each action is a module that can satisfy the information needs of users for some requests. For example, search and question answering can be two actions in Macaw. Even multiple search algorithms can be also seen as multiple actions. Each action can produce multiple outputs (e.g., multiple retrieved documents). For every user interaction, Macaw runs all actions in parallel. The actions' outputs produced within a predefined time interval (i.e., an interaction timeout constant) are then post-processed. Macaw can choose one or combine multiple of these outputs and prepare an output Message object as the user's response.
The modular design of Macaw makes it relatively easy to configure a different user interface or add a new one. The current implementation of Macaw supports a command line interface as well as mobile, desktop, and web apps. In more detail, Macaw's interface can be a Telegram bot, which supports a wide range of devices and operating systems (see FIGREF4). This allows Macaw to support multi-modal interactions, such as text, speech, image, click, etc. A number of APIs for automatic speech recognition and generation have been employed to support speech interactions. Note that the Macaw's architecture and implementation allows mixed-initiative interactions.
The research community can benefit from Macaw for the following purposes:
[leftmargin=*]
Developing algorithms, tools, and techniques for CIS.
Studying user interactions with CIS systems.
Performing CIS studies based on an intermediary person and wizard of oz.
Preparing quick demonstration for a developed CIS model.
Macaw Architecture
Macaw has a modular design, with the goal of making it easy to configure and add new modules such as a different user interface or different retrieval module. The overall setup also follows a Model-View-Controller (MVC) like architecture. The design decisions have been made to smooth the Macaw's adoptions and extensions. Macaw is implemented in Python, thus machine learning models implemented using PyTorch, Scikit-learn, or TensorFlow can be easily integrated into Macaw. The high-level overview of Macaw is depicted in FIGREF8. The user interacts with the interface and the interface produces a Message object from the current interaction of user. The interaction can be in multi-modal form, such as text, speech, image, and click. Macaw stores all interactions in an “Interaction Database”. For every interaction, Macaw looks for most recent user-system interactions (including the system's responses) to create a list of Messages, called the conversation list. It is then dispatched to multiple information seeking (and related) actions. The actions run in parallel, and each should respond within a pre-defined time interval. The output selection component selects from (or potentially combines) the outputs generated by different actions and creates a Message object as the system's response. This message is logged into the interaction database and is sent to the interface to be presented to the user. Again, the response message can be multi-modal and include text, speech, link, list of options, etc.
Macaw also supports Wizard of Oz studies or intermediary-based information seeking studies. The architecture of Macaw for such setup is presented in FIGREF16. As shown in the figure, the seeker interacts with a real conversational interface that supports multi-modal and mixed-initiative interactions in multiple devices. The intermediary (or the wizard) receives the seeker's message and performs different information seeking actions with Macaw. All seeker-intermediary and intermediary-system interactions will be logged for further analysis. This setup can simulate an ideal CIS system and thus is useful for collecting high-quality data from real users for CIS research.
Retrieval and Question Answering in Macaw
The overview of retrieval and question answering actions in Macaw is shown in FIGREF17. These actions consist of the following components:
[leftmargin=*]
Co-Reference Resolution: To support multi-turn interactions, it is sometimes necessary to use co-reference resolution techniques for effective retrieval. In Macaw, we identify all the co-references from the last request of user to the conversation history. The same co-reference resolution outputs can be used for different query generation components. This can be a generic or action-specific component.
Query Generation: This component generates a query based on the past user-system interactions. The query generation component may take advantage of co-reference resolution for query expansion or re-writing.
Retrieval Model: This is the core ranking component that retrieves documents or passages from a large collection. Macaw can retrieve documents from an arbitrary document collection using the Indri python interface BIBREF9, BIBREF10. We also provide the support for web search using the Bing Web Search API. Macaw also allows multi-stage document re-ranking.
Result Generation: The retrieved documents can be too long to be presented using some interfaces. Result generation is basically a post-processing step ran on the retrieved result list. In case of question answering, it can employ answer selection or generation techniques, such as machine reading comprehension models. For example, Macaw features the DrQA model BIBREF11 for question answering.
These components are implemented in a generic form, so researchers can easily replace them with their own favorite algorithms.
User Interfaces
We have implemented the following interfaces for Macaw:
[leftmargin=*]
File IO: This interface is designed for experimental purposes, such as evaluating the performance of a conversational search technique on a dataset with multiple queries. This is not an interactive interface.
Standard IO: This interactive command line interface is designed for development purposes to interact with the system, see the logs, and debug or improve the system.
Telegram: This interactive interface is designed for interaction with real users (see FIGREF4). Telegram is a popular instant messaging service whose client-side code is open-source. We have implemented a Telegram bot that can be used with different devices (personal computers, tablets, and mobile phones) and different operating systems (Android, iOS, Linux, Mac OS, and Windows). This interface allows multi-modal interactions (text, speech, click, image). It can be also used for speech-only interactions. For speech recognition and generation, Macaw relies on online APIs, e.g., the services provided by Google Cloud and Microsoft Azure. In addition, there exist multiple popular groups and channels in Telegram, which allows further integration of social networks with conversational systems. For example, see the Naseri and Zamani's study on news popularity in Telegram BIBREF12.
Similar to the other modules, one can easily extend Macaw using other appropriate user interfaces.
Limitations and Future Work
The current implementation of Macaw lacks the following actions. We intend to incrementally improve Macaw by supporting more actions and even more advanced techniques for the developed actions.
[leftmargin=*]
Clarification and Preference Elicitation: Asking clarifying questions has been recently recognized as a necessary component in a conversational system BIBREF3, BIBREF7. The authors are not aware of a published solution for generating clarifying questions using public resources. Therefore, Macaw does not currently support clarification.
Explanation: Despite its importance, result list explanation is also a relatively less explored topic. We intend to extend Macaw with result list explanation as soon as we find a stable and mature solution.
Recommendation: In our first release, we focus on conversational search and question answering tasks. We intend to provide support for conversational recommendation, e.g., BIBREF13, BIBREF14, BIBREF15, and joint search and recommendation, e.g., BIBREF16, BIBREF17, in the future.
Natural Language Interface: Macaw can potentially support access to structured data, such as knowledge graph. We would like to ease conversational natural language interface to structured and semi-structured data in our future releases.
Contribution
Macaw is distributed under the MIT License. We welcome contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com. This project has adopted the Microsoft Open Source Code of Conduct.
When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.
Conclusions
This paper described Macaw, an open-source platform for conversational information seeking research. Macaw supports multi-turn, multi-modal, and mixed-initiative interactions. It was designed based on a modular architecture that allows further improvements and extensions. Researchers can benefit from Macaw for developing algorithms and techniques for conversational information seeking research, for user studies with different interfaces, for data collection from real users, and for preparing a demonstration of a CIS model.
Acknowledgements
The authors wish to thank Ahmed Hassan Awadallah, Krisztian Balog, and Arjen P. de Vries for their invaluable feedback.
Question:
Does the paper provide any case studies to illustrate how one can use Macaw for CIS research?
Answer:
| No | 328 | 11,979 | 11,981 | 12,084 | ... [The rest of the article is omitted]
|
290ee79b5e3872e0496a6a0fc9b103ab7d8f6c30 | 290ee79b5e3872e0496a6a0fc9b103ab7d8f6c30_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Offensive content has become pervasive in social media and a reason of concern for government organizations, online communities, and social media platforms. One of the most common strategies to tackle the problem is to train systems capable of recognizing offensive content, which then can be deleted or set aside for human moderation. In the last few years, there have been several studies published on the application of computational methods to deal with this problem. Most prior work focuses on a different aspect of offensive language such as abusive language BIBREF0 , BIBREF1 , (cyber-)aggression BIBREF2 , (cyber-)bullying BIBREF3 , BIBREF4 , toxic comments INLINEFORM0 , hate speech BIBREF5 , BIBREF6 , BIBREF7 , BIBREF8 , BIBREF9 , BIBREF10 , and offensive language BIBREF11 . Prior work has focused on these aspects of offensive language in Twitter BIBREF3 , BIBREF7 , BIBREF8 , BIBREF11 , Wikipedia comments, and Facebook posts BIBREF2 .
Recently, Waseem et. al. ( BIBREF12 ) acknowledged the similarities among prior work and discussed the need for a typology that differentiates between whether the (abusive) language is directed towards a specific individual or entity or towards a generalized group and whether the abusive content is explicit or implicit. Wiegand et al. ( BIBREF11 ) followed this trend as well on German tweets. In their evaluation, they have a task to detect offensive vs not offensive tweets and a second task for distinguishing between the offensive tweets as profanity, insult, or abuse. However, no prior work has explored the target of the offensive language, which is important in many scenarios, e.g., when studying hate speech with respect to a specific target.
Therefore, we expand on these ideas by proposing a a hierarchical three-level annotation model that encompasses:
Using this annotation model, we create a new large publicly available dataset of English tweets. The key contributions of this paper are as follows:
Related Work
Different abusive and offense language identification sub-tasks have been explored in the past few years including aggression identification, bullying detection, hate speech, toxic comments, and offensive language.
Aggression identification: The TRAC shared task on Aggression Identification BIBREF2 provided participants with a dataset containing 15,000 annotated Facebook posts and comments in English and Hindi for training and validation. For testing, two different sets, one from Facebook and one from Twitter were provided. Systems were trained to discriminate between three classes: non-aggressive, covertly aggressive, and overtly aggressive.
Bullying detection: Several studies have been published on bullying detection. One of them is the one by xu2012learning which apply sentiment analysis to detect bullying in tweets. xu2012learning use topic models to to identify relevant topics in bullying. Another related study is the one by dadvar2013improving which use user-related features such as the frequency of profanity in previous messages to improve bullying detection.
Hate speech identification: It is perhaps the most widespread abusive language detection sub-task. There have been several studies published on this sub-task such as kwok2013locate and djuric2015hate who build a binary classifier to distinguish between `clean' comments and comments containing hate speech and profanity. More recently, Davidson et al. davidson2017automated presented the hate speech detection dataset containing over 24,000 English tweets labeled as non offensive, hate speech, and profanity.
Offensive language: The GermEval BIBREF11 shared task focused on Offensive language identification in German tweets. A dataset of over 8,500 annotated tweets was provided for a course-grained binary classification task in which systems were trained to discriminate between offensive and non-offensive tweets and a second task where the organizers broke down the offensive class into three classes: profanity, insult, and abuse.
Toxic comments: The Toxic Comment Classification Challenge was an open competition at Kaggle which provided participants with comments from Wikipedia labeled in six classes: toxic, severe toxic, obscene, threat, insult, identity hate.
While each of these sub-tasks tackle a particular type of abuse or offense, they share similar properties and the hierarchical annotation model proposed in this paper aims to capture this. Considering that, for example, an insult targeted at an individual is commonly known as cyberbulling and that insults targeted at a group are known as hate speech, we pose that OLID's hierarchical annotation model makes it a useful resource for various offensive language identification sub-tasks.
Hierarchically Modelling Offensive Content
In the OLID dataset, we use a hierarchical annotation model split into three levels to distinguish between whether language is offensive or not (A), and type (B) and target (C) of the offensive language. Each level is described in more detail in the following subsections and examples are shown in Table TABREF10 .
Level A: Offensive language Detection
Level A discriminates between offensive (OFF) and non-offensive (NOT) tweets.
Not Offensive (NOT): Posts that do not contain offense or profanity;
Offensive (OFF): We label a post as offensive if it contains any form of non-acceptable language (profanity) or a targeted offense, which can be veiled or direct. This category includes insults, threats, and posts containing profane language or swear words.
Level B: Categorization of Offensive Language
Level B categorizes the type of offense and two labels are used: targeted (TIN) and untargeted (INT) insults and threats.
Targeted Insult (TIN): Posts which contain an insult/threat to an individual, group, or others (see next layer);
Untargeted (UNT): Posts containing non-targeted profanity and swearing. Posts with general profanity are not targeted, but they contain non-acceptable language.
Level C: Offensive Language Target Identification
Level C categorizes the targets of insults and threats as individual (IND), group (GRP), and other (OTH).
Individual (IND): Posts targeting an individual. It can be a a famous person, a named individual or an unnamed participant in the conversation. Insults and threats targeted at individuals are often defined as cyberbulling.
Group (GRP): The target of these offensive posts is a group of people considered as a unity due to the same ethnicity, gender or sexual orientation, political affiliation, religious belief, or other common characteristic. Many of the insults and threats targeted at a group correspond to what is commonly understood as hate speech.
Other (OTH): The target of these offensive posts does not belong to any of the previous two categories (e.g. an organization, a situation, an event, or an issue).
Data Collection
The data included in OLID has been collected from Twitter. We retrieved the data using the Twitter API by searching for keywords and constructions that are often included in offensive messages, such as `she is' or `to:BreitBartNews'. We carried out a first round of trial annotation of 300 instances with six experts. The goal of the trial annotation was to 1) evaluate the proposed tagset; 2) evaluate the data retrieval method; and 3) create a gold standard with instances that could be used as test questions in the training and test setting annotation which was carried out using crowdsourcing. The breakdown of keywords and their offensive content in the trial data of 300 tweets is shown in Table TABREF14 . We included a left (@NewYorker) and far-right (@BreitBartNews) news accounts because there tends to be political offense in the comments. One of the best offensive keywords was tweets that were flagged as not being safe by the Twitter `safe' filter (the `-' indicates `not safe'). The vast majority of content on Twitter is not offensive so we tried different strategies to keep a reasonable number of tweets in the offensive class amounting to around 30% of the dataset including excluding some keywords that were not high in offensive content such as `they are` and `to:NewYorker`. Although `he is' is lower in offensive content we kept it as a keyword to avoid gender bias. In addition to the keywords in the trial set, we searched for more political keywords which tend to be higher in offensive content, and sampled our dataset such that 50% of the the tweets come from political keywords and 50% come from non-political keywords. In addition to the keywords `gun control', and `to:BreitbartNews', political keywords used to collect these tweets are `MAGA', `antifa', `conservative' and `liberal'. We computed Fliess' INLINEFORM0 on the trial set for the five annotators on 21 of the tweets. INLINEFORM1 is .83 for Layer A (OFF vs NOT) indicating high agreement. As to normalization and anonymization, no user metadata or Twitter IDs have been stored, and URLs and Twitter mentions have been substituted to placeholders. We follow prior work in related areas (burnap2015cyber,davidson2017automated) and annotate our data using crowdsourcing using the platform Figure Eight. We ensure data quality by: 1) we only received annotations from individuals who were experienced in the platform; and 2) we used test questions to discard annotations of individuals who did not reach a certain threshold. Each instance in the dataset was annotated by multiple annotators and inter-annotator agreement has been calculated. We first acquired two annotations for each instance. In case of 100% agreement, we considered these as acceptable annotations, and in case of disagreement, we requested more annotations until the agreement was above 66%. After the crowdsourcing annotation, we used expert adjudication to guarantee the quality of the annotation. The breakdown of the data into training and testing for the labels from each level is shown in Table TABREF15 .
Experiments and Evaluation
We assess our dataset using traditional and deep learning methods. Our simplest model is a linear SVM trained on word unigrams. SVMs have produced state-of-the-art results for many text classification tasks BIBREF13 . We also train a bidirectional Long Short-Term-Memory (BiLSTM) model, which we adapted from the sentiment analysis system of sentimentSystem,rasooli2018cross and altered to predict offensive labels instead. It consists of (1) an input embedding layer, (2) a bidirectional LSTM layer, (3) an average pooling layer of input features. The concatenation of the LSTM's and average pool layer is passed through a dense layer and the output is passed through a softmax function. We set two input channels for the input embedding layers: pre-trained FastText embeddings BIBREF14 , as well as updatable embeddings learned by the model during training. Finally, we also apply a Convolutional Neural Network (CNN) model based on the architecture of BIBREF15 , using the same multi-channel inputs as the above BiLSTM.
Our models are trained on the training data, and evaluated by predicting the labels for the held-out test set. The distribution is described in Table TABREF15 . We evaluate and compare the models using the macro-averaged F1-score as the label distribution is highly imbalanced. Per-class Precision (P), Recall (R), and F1-score (F1), also with other averaged metrics are also reported. The models are compared against baselines of predicting all labels as the majority or minority classes.
Offensive Language Detection
The performance on discriminating between offensive (OFF) and non-offensive (NOT) posts is reported in Table TABREF18 . We can see that all systems perform significantly better than chance, with the neural models being substantially better than the SVM. The CNN outperforms the RNN model, achieving a macro-F1 score of 0.80.
Categorization of Offensive Language
In this experiment, the two systems were trained to discriminate between insults and threats (TIN) and untargeted (UNT) offenses, which generally refer to profanity. The results are shown in Table TABREF19 .
The CNN system achieved higher performance in this experiment compared to the BiLSTM, with a macro-F1 score of 0.69. All systems performed better at identifying target and threats (TIN) than untargeted offenses (UNT).
Offensive Language Target Identification
The results of the offensive target identification experiment are reported in Table TABREF20 . Here the systems were trained to distinguish between three targets: a group (GRP), an individual (IND), or others (OTH). All three models achieved similar results far surpassing the random baselines, with a slight performance edge for the neural models.
The performance of all systems for the OTH class is 0. This poor performances can be explained by two main factors. First, unlike the two other classes, OTH is a heterogeneous collection of targets. It includes offensive tweets targeted at organizations, situations, events, etc. making it more challenging for systems to learn discriminative properties of this class. Second, this class contains fewer training instances than the other two. There are only 395 instances in OTH, and 1,075 in GRP, and 2,407 in IND.
Conclusion and Future Work
This paper presents OLID, a new dataset with annotation of type and target of offensive language. OLID is the official dataset of the shared task SemEval 2019 Task 6: Identifying and Categorizing Offensive Language in Social Media (OffensEval) BIBREF16 . In OffensEval, each annotation level in OLID is an independent sub-task. The dataset contains 14,100 tweets and is released freely to the research community. To the best of our knowledge, this is the first dataset to contain annotation of type and target of offenses in social media and it opens several new avenues for research in this area. We present baseline experiments using SVMs and neural networks to identify the offensive tweets, discriminate between insults, threats, and profanity, and finally to identify the target of the offensive messages. The results show that this is a challenging task. A CNN-based sentence classifier achieved the best results in all three sub-tasks.
In future work, we would like to make a cross-corpus comparison of OLID and datasets annotated for similar tasks such as aggression identification BIBREF2 and hate speech detection BIBREF8 . This comparison is, however, far from trivial as the annotation of OLID is different.
Acknowledgments
The research presented in this paper was partially supported by an ERAS fellowship awarded by the University of Wolverhampton.
Question:
What are the three layers of the annotation scheme?
Answer:
| Level A: Offensive language Detection
, Level B: Categorization of Offensive Language
, Level C: Offensive Language Target Identification
| 328 | 14,920 | 14,922 | 14,983 | ... [The rest of the article is omitted]
|
ab9b0bde6113ffef8eb1c39919d21e5913a05081 | ab9b0bde6113ffef8eb1c39919d21e5913a05081_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Writing errors can occur in many different forms – from relatively simple punctuation and determiner errors, to mistakes including word tense and form, incorrect collocations and erroneous idioms. Automatically identifying all of these errors is a challenging task, especially as the amount of available annotated data is very limited. Rei2016 showed that while some error detection algorithms perform better than others, it is additional training data that has the biggest impact on improving performance.
Being able to generate realistic artificial data would allow for any grammatically correct text to be transformed into annotated examples containing writing errors, producing large amounts of additional training examples. Supervised error generation systems would also provide an efficient method for anonymising the source corpus – error statistics from a private corpus can be aggregated and applied to a different target text, obscuring sensitive information in the original examination scripts. However, the task of creating incorrect data is somewhat more difficult than might initially appear – naive methods for error generation can create data that does not resemble natural errors, thereby making downstream systems learn misleading or uninformative patterns.
Previous work on artificial error generation (AEG) has focused on specific error types, such as prepositions and determiners BIBREF0 , BIBREF1 , or noun number errors BIBREF2 . Felice2014a investigated the use of linguistic information when generating artificial data for error correction, but also restricting the approach to only five error types. There has been very limited research on generating artificial data for all types, which is important for general-purpose error detection systems. For example, the error types investigated by Felice2014a cover only 35.74% of all errors present in the CoNLL 2014 training dataset, providing no additional information for the majority of errors.
In this paper, we investigate two supervised approaches for generating all types of artificial errors. We propose a framework for generating errors based on statistical machine translation (SMT), training a model to translate from correct into incorrect sentences. In addition, we describe a method for learning error patterns from an annotated corpus and transplanting them into error-free text. We evaluate the effect of introducing artificial data on two error detection benchmarks. Our results show that each method provides significant improvements over using only the available training set, and a combination of both gives an absolute improvement of 4.3% in INLINEFORM0 , without requiring any additional annotated data.
Error Generation Methods
We investigate two alternative methods for AEG. The models receive grammatically correct text as input and modify certain tokens to produce incorrect sequences. The alternative versions of each sentence are aligned using Levenshtein distance, allowing us to identify specific words that need to be marked as errors. While these alignments are not always perfect, we found them to be sufficient for practical purposes, since alternative alignments of similar sentences often result in the same binary labeling. Future work could explore more advanced alignment methods, such as proposed by felice-bryant-briscoe.
In Section SECREF4 , this automatically labeled data is then used for training error detection models.
Machine Translation
We treat AEG as a translation task – given a correct sentence as input, the system would learn to translate it to contain likely errors, based on a training corpus of parallel data. Existing SMT approaches are already optimised for identifying context patterns that correspond to specific output sequences, which is also required for generating human-like errors. The reverse of this idea, translating from incorrect to correct sentences, has been shown to work well for error correction tasks BIBREF2 , BIBREF3 , and round-trip translation has also been shown to be promising for correcting grammatical errors BIBREF4 .
Following previous work BIBREF2 , BIBREF5 , we build a phrase-based SMT error generation system. During training, error-corrected sentences in the training data are treated as the source, and the original sentences written by language learners as the target. Pialign BIBREF6 is used to create a phrase translation table directly from model probabilities. In addition to default features, we add character-level Levenshtein distance to each mapping in the phrase table, as proposed by Felice:2014-CoNLL. Decoding is performed using Moses BIBREF7 and the language model used during decoding is built from the original erroneous sentences in the learner corpus. The IRSTLM Toolkit BIBREF8 is used for building a 5-gram language model with modified Kneser-Ney smoothing BIBREF9 .
Pattern Extraction
We also describe a method for AEG using patterns over words and part-of-speech (POS) tags, extracting known incorrect sequences from a corpus of annotated corrections. This approach is based on the best method identified by Felice2014a, using error type distributions; while they covered only 5 error types, we relax this restriction and learn patterns for generating all types of errors.
The original and corrected sentences in the corpus are aligned and used to identify short transformation patterns in the form of (incorrect phrase, correct phrase). The length of each pattern is the affected phrase, plus up to one token of context on both sides. If a word form changes between the incorrect and correct text, it is fully saved in the pattern, otherwise the POS tags are used for matching.
For example, the original sentence `We went shop on Saturday' and the corrected version `We went shopping on Saturday' would produce the following pattern:
(VVD shop_VV0 II, VVD shopping_VVG II)
After collecting statistics from the background corpus, errors can be inserted into error-free text. The learned patterns are now reversed, looking for the correct side of the tuple in the input sentence. We only use patterns with frequency INLINEFORM0 , which yields a total of 35,625 patterns from our training data. For each input sentence, we first decide how many errors will be generated (using probabilities from the background corpus) and attempt to create them by sampling from the collection of applicable patterns. This process is repeated until all the required errors have been generated or the sentence is exhausted. During generation, we try to balance the distribution of error types as well as keeping the same proportion of incorrect and correct sentences as in the background corpus BIBREF10 . The required POS tags were generated with RASP BIBREF11 , using the CLAWS2 tagset.
Error Detection Model
We construct a neural sequence labeling model for error detection, following the previous work BIBREF12 , BIBREF13 . The model receives a sequence of tokens as input and outputs a prediction for each position, indicating whether the token is correct or incorrect in the current context. The tokens are first mapped to a distributed vector space, resulting in a sequence of word embeddings. Next, the embeddings are given as input to a bidirectional LSTM BIBREF14 , in order to create context-dependent representations for every token. The hidden states from forward- and backward-LSTMs are concatenated for each word position, resulting in representations that are conditioned on the whole sequence. This concatenated vector is then passed through an additional feedforward layer, and a softmax over the two possible labels (correct and incorrect) is used to output a probability distribution for each token. The model is optimised by minimising categorical cross-entropy with respect to the correct labels. We use AdaDelta BIBREF15 for calculating an adaptive learning rate during training, which accounts for a higher baseline performance compared to previous results.
Evaluation
We trained our error generation models on the public FCE training set BIBREF16 and used them to generate additional artificial training data. Grammatically correct text is needed as the starting point for inserting artificial errors, and we used two different sources: 1) the corrected version of the same FCE training set on which the system is trained (450K tokens), and 2) example sentences extracted from the English Vocabulary Profile (270K tokens).. While there are other text corpora that could be used (e.g., Wikipedia and news articles), our development experiments showed that keeping the writing style and vocabulary close to the target domain gives better results compared to simply including more data.
We evaluated our detection models on three benchmarks: the FCE test data (41K tokens) and the two alternative annotations of the CoNLL 2014 Shared Task dataset (30K tokens) BIBREF3 . Each artificial error generation system was used to generate 3 different versions of the artificial data, which were then combined with the original annotated dataset and used for training an error detection system. Table TABREF1 contains example sentences from the error generation systems, highlighting each of the edits that are marked as errors.
The error detection results can be seen in Table TABREF4 . We use INLINEFORM0 as the main evaluation measure, which was established as the preferred measure for error correction and detection by the CoNLL-14 shared task BIBREF3 . INLINEFORM1 calculates a weighted harmonic mean of precision and recall, which assigns twice as much importance to precision – this is motivated by practical applications, where accurate predictions from an error detection system are more important compared to coverage. For comparison, we also report the performance of the error detection system by Rei2016, trained using the same FCE dataset.
The results show that error detection performance is substantially improved by making use of artificially generated data, created by any of the described methods. When comparing the error generation system by Felice2014a (FY14) with our pattern-based (PAT) and machine translation (MT) approaches, we see that the latter methods covering all error types consistently improve performance. While the added error types tend to be less frequent and more complicated to capture, the added coverage is indeed beneficial for error detection. Combining the pattern-based approach with the machine translation system (Ann+PAT+MT) gave the best overall performance on all datasets. The two frameworks learn to generate different types of errors, and taking advantage of both leads to substantial improvements in error detection.
We used the Approximate Randomisation Test BIBREF17 , BIBREF18 to calculate statistical significance and found that the improvement for each of the systems using artificial data was significant over using only manual annotation. In addition, the final combination system is also significantly better compared to the Felice2014a system, on all three datasets. While Rei2016 also report separate experiments that achieve even higher performance, these models were trained on a considerably larger proprietary corpus. In this paper we compare error detection frameworks trained on the same publicly available FCE dataset, thereby removing the confounding factor of dataset size and only focusing on the model architectures.
The error generation methods can generate alternative versions of the same input text – the pattern-based method randomly samples the error locations, and the SMT system can provide an n-best list of alternative translations. Therefore, we also investigated the combination of multiple error-generated versions of the input files when training error detection models. Figure FIGREF6 shows the INLINEFORM0 score on the development set, as the training data is increased by using more translations from the n-best list of the SMT system. These results reveal that allowing the model to see multiple alternative versions of the same file gives a distinct improvement – showing the model both correct and incorrect variations of the same sentences likely assists in learning a discriminative model.
Related Work
Our work builds on prior research into AEG. Brockett2006 constructed regular expressions for transforming correct sentences to contain noun number errors. Rozovskaya2010a learned confusion sets from an annotated corpus in order to generate preposition errors. Foster2009 devised a tool for generating errors for different types using patterns provided by the user or collected automatically from an annotated corpus. However, their method uses a limited number of edit operations and is thus unable to generate complex errors. Cahill2013 compared different training methodologies and showed that artificial errors helped correct prepositions. Felice2014a learned error type distributions for generating five types of errors, and the system in Section SECREF3 is an extension of this model. While previous work focused on generating a specific subset of error types, we explored two holistic approaches to AEG and showed that they are able to significantly improve error detection performance.
Conclusion
This paper investigated two AEG methods, in order to create additional training data for error detection. First, we explored a method using textual patterns learned from an annotated corpus, which are used for inserting errors into correct input text. In addition, we proposed formulating error generation as an MT framework, learning to translate from grammatically correct to incorrect sentences.
The addition of artificial data to the training process was evaluated on three error detection annotations, using the FCE and CoNLL 2014 datasets. Making use of artificial data provided improvements for all data generation methods. By relaxing the type restrictions and generating all types of errors, our pattern-based method consistently outperformed the system by Felice2014a. The combination of the pattern-based method with the machine translation approach gave further substantial improvements and the best performance on all datasets.
Question:
What are their results on both datasets?
Answer:
| Combining pattern based and Machine translation approaches gave the best overall F0.5 scores. It was 49.11 for FCE dataset , 21.87 for the first annotation of CoNLL-14, and 30.13 for the second annotation of CoNLL-14. | 328 | 14,471 | 14,473 | 14,523 | ... [The rest of the article is omitted]
|
ff338921e34c15baf1eae0074938bf79ee65fdd2 | ff338921e34c15baf1eae0074938bf79ee65fdd2_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
BioASQ is a biomedical document classification, document retrieval, and question answering competition, currently in its seventh year. We provide an overview of our submissions to semantic question answering task (7b, Phase B) of BioASQ 7 (except for 'ideal answer' test, in which we did not participate this year). In this task systems are provided with biomedical questions and are required to submit ideal and exact answers to those questions. We have used BioBERT BIBREF0 based system , see also Bidirectional Encoder Representations from Transformers(BERT) BIBREF1, and we fine tuned it for the biomedical question answering task. Our system scored near the top for factoid questions for all the batches of the challenge. More specifially, in the third test batch set, our system achieved highest ‘MRR’ score for Factoid Question Answering task. Also, for List-type question answering task our system achieved highest recall score in the fourth test batch set. Along with our detailed approach, we present the results for our submissions and also highlight identified downsides for our current approach and ways to improve them in our future experiments. In last test batch results we placed 4th for List-type questions and 3rd for Factoid-type questions.)
The QA task is organized in two phases. Phase A deals with retrieval of the relevant document, snippets, concepts, and RDF triples, and phase B deals with exact and ideal answer generations (which is a paragraph size summary of snippets). Exact answer generation is required for factoid, list, and yes/no type question.
BioASQ organizers provide the training and testing data. The training data consists of questions, gold standard documents, snippets, concepts, and ideal answers (which we did not use in this paper, but we used last year BIBREF2). The test data is split between phases A and B. The Phase A dataset consists of the questions, unique ids, question types. The Phase B dataset consists of the questions, golden standard documents, snippets, unique ids and question types. Exact answers for factoid type questions are evaluated using strict accuracy (the top answer), lenient accuracy (the top 5 answers), and MRR (Mean Reciprocal Rank) which takes into account the ranks of returned answers. Answers for the list type question are evaluated using precision, recall, and F-measure.
Related Work ::: BioAsq
Sharma et al. BIBREF3 describe a system with two stage process for factoid and list type question answering. Their system extracts relevant entities and then runs supervised classifier to rank the entities. Wiese et al. BIBREF4 propose neural network based model for Factoid and List-type question answering task. The model is based on Fast QA and predicts the answer span in the passage for a given question. The model is trained on SQuAD data set and fine tuned on the BioASQ data. Dimitriadis et al. BIBREF5 proposed two stage process for Factoid question answering task. Their system uses general purpose tools such as Metamap, BeCas to identify candidate sentences. These candidate sentences are represented in the form of features, and are then ranked by the binary classifier. Classifier is trained on candidate sentences extracted from relevant questions, snippets and correct answers from BioASQ challenge. For factoid question answering task highest ‘MRR’ achieved in the 6th edition of BioASQ competition is ‘0.4325’. Our system is a neural network model based on contextual word embeddings BIBREF1 and achieved a ‘MRR’ score ‘0.6103’ in one of the test batches for Factoid Question Answering task.
Related Work ::: A minimum background on BERT
BERT stands for "Bidirectional Encoder Representations from Transformers" BIBREF1 is a contextual word embedding model. Given a sentence as an input, contextual embedding for the words are returned. The BERT model was designed so it can be fine tuned for 11 different tasks BIBREF1, including question answering tasks. For a question answering task, question and paragraph (context) are given as an input. A BERT standard is that question text and paragraph text are separated by a separator [Sep]. BERT question-answering fine tuning involves adding softmax layer. Softmax layer takes contextual word embeddings from BERT as input and learns to identity answer span present in the paragraph (context). This process is represented in Figure FIGREF4.
BERT was originally trained to perform tasks such as language model creation using masked words and next-sentence-prediction. In other words BERT weights are learned such that context is used in building the representation of the word, not just as a loss function to help learn a context-independent representation. For detailed understanding of BERT Architecture, please refer to the original BERT paper BIBREF1.
Related Work ::: A minimum background on BERT ::: Comparison of Word Embeddings and Contextual Word Embeddings
A ‘word embedding’ is a learned representation. It is represented in the form of vector where words that have the same meaning have a similar vector representation. Consider a word embedding model 'word2vec' BIBREF6 trained on a corpus. Word embeddings generated from the model are context independent that is, word embeddings are returned regardless of where the words appear in a sentence and regardless of e.g. the sentiment of the sentence. However, contextual word embedding models like BERT also takes context of the word into consideration.
Related Work ::: Comparison of BERT and Bio-BERT
‘BERT’ and BioBERT are very similar in terms of architecture. Difference is that ‘BERT’ is pretrained on Wikipedia articles, whereas BioBERT version used in our experiments is pretrained on Wikipedia, PMC and PubMed articles. Therefore BioBERT model is expected to perform well with biomedical text, in terms of generating contextual word embeddings.
BioBERT model used in our experiments is based on BERT-Base Architecture; BERT-Base has 12 transformer Layers where as BERT-Large has 24 transformer layers. Moreover contextual word embedding vector size is 768 for BERT-Base and more for BERT-large. According to BIBREF1 Bert-Large, fine-tuned on SQuAD 1.1 question answering data BIBREF7 can achieve F1 Score of 90.9 for Question Answering task where as BERT-Base Fine-tuned on the same SQuAD question answering BIBREF7 data could achieve F1 score of 88.5. One downside of the current version BioBERT is that word-piece vocabulary is the same as that of original BERT Model, as a result word-piece vocabulary does not include biomedical jargon. Lee et al. BIBREF0 created BioBERT, using the same pre-trained BERT released by Google, and hence in the word-piece vocabulary (vocab.txt), as a result biomedical jargon is not included in word-piece vocabulary. Modifying word-piece vocabulary (vocab.txt) at this stage would loose original compatibility with ‘BERT’, hence it is left unmodified.
In our future work we would like to build pre-trained ‘BERT’ model from scratch. We would pretrain the model with biomedical corpus (PubMed, ‘PMC’) and Wikipedia. Doing so would give us scope to create word piece vocabulary to include biomedical jargon and there are chances of model performing better with biomedical jargon being included in the word piece vocabulary. We will consider this scenario in the future, or wait for the next version of BioBERT.
Experiments: Factoid Question Answering Task
For Factoid Question Answering task, we fine tuned BioBERT BIBREF0 with question answering data and added new features. Fig. FIGREF4 shows the architecture of BioBERT fine tuned for question answering tasks: Input to BioBERT is word tokenized embeddings for question and the paragraph (Context). As per the ‘BERT’ BIBREF1 standards, tokens ‘[CLS]’ and ‘[SEP]’ are appended to the tokenized input as illustrated in the figure. The resulting model has a softmax layer formed for predicting answer span indices in the given paragraph (Context). On test data, the fine tuned model generates $n$-best predictions for each question. For a question, $n$-best corresponds that $n$ answers are returned as possible answers in the decreasing order of confidence. Variable $n$ is configurable. In our paper, any further mentions of ‘answer returned by the model’ correspond to the top answer returned by the model.
Experiments: Factoid Question Answering Task ::: Setup
BioASQ provides the training data. This data is based on previous BioASQ competitions. Train data we have considered is aggregate of all train data sets till the 5th version of BioASQ competition. We cleaned the data, that is, question-answering data without answers are removed and left with a total count of ‘530’ question answers. The data is split into train and test data in the ratio of 94 to 6; that is, count of '495' for training and '35' for testing.
The original data format is converted to the BERT/BioBERT format, where BioBERT expects ‘start_index’ of the actual answer. The ‘start_index corresponds to the index of the answer text present in the paragraph/ Context. For finding ‘start_index’ we used built-in python function find(). The function returns the lowest index of the actual answer present in the context(paragraph). If the answer is not found ‘-1’ is returned as the index. The efficient way of finding start_index is, if the paragraph (Context) has multiple instances of answer text, then ‘start_index’ of the answer should be that instance of answer text whose context actually matches with what’s been asked in the question.
Example (Question, Answer and Paragraph from BIBREF8):
Question: Which drug should be used as an antidote in benzodiazepine overdose?
Answer: 'Flumazenil'
Paragraph(context):
"Flumazenil use in benzodiazepine overdose in the UK: a retrospective survey of NPIS data. OBJECTIVE: Benzodiazepine (BZD) overdose (OD) continues to cause significant morbidity and mortality in the UK. Flumazenil is an effective antidote but there is a risk of seizures, particularly in those who have co-ingested tricyclic antidepressants. A study was undertaken to examine the frequency of use, safety and efficacy of flumazenil in the management of BZD OD in the UK. METHODS: A 2-year retrospective cohort study was performed of all enquiries to the UK National Poisons Information Service involving BZD OD. RESULTS: Flumazenil was administered to 80 patients in 4504 BZD-related enquiries, 68 of whom did not have ventilatory failure or had recognised contraindications to flumazenil. Factors associated with flumazenil use were increased age, severe poisoning and ventilatory failure. Co-ingestion of tricyclic antidepressants and chronic obstructive pulmonary disease did not influence flumazenil administration. Seizure frequency in patients not treated with flumazenil was 0.3%".
Actual answer is 'Flumazenil', but there are multiple instances of word 'Flu-mazenil'. Efficient way to identify the start-index for 'Flumazenil'(answer) is to find that particular instance of the word 'Flumazenil' which matches the context for the question. In the above example 'Flumazenil' highlighted in bold is the actual instance that matches question's context. Unfortunately, we could not identify readily available tools that can achieve this goal. In our future work, we look forward to handling these scenarios effectively.
Note: The creators of 'SQuAD' BIBREF7 have handled the task of identifying answer's start_index effectively. But 'SQuAD' data set is much more general and does not include biomedical question answering data.
Experiments: Factoid Question Answering Task ::: Training and error analysis
During our training with the BioASQ data, learning rate is set to 3e-5, as mentioned in the BioBERT paper BIBREF0. We started training the model with 495 available train data and 35 test data by setting number of epochs to 50. After training with these hyper-parameters training accuracy(exact match) was 99.3%(overfitting) and testing accuracy is only 4%. In the next iteration we reduced the number of epochs to 25 then training accuracy is reduced to 98.5% and test accuracy moved to 5%. We further reduced number of epochs to 15, and the resulting training accuracy was 70% and test accuracy 15%. In the next iteration set number of epochs to 12 and achieved train accuracy of 57.7% and test accuracy 23.3%. Repeated the experiment with 11 epochs and found training accuracy to be 57.7% and test accuracy to be same 22%. In the next iteration we set number of epochs to '9' and found training accuracy of 48% and test accuracy of 15%. Hence optimum number of epochs is taken as 12 epochs.
During our error analysis we found that on test data, model tends to return text in the beginning of the context(paragraph) as the answer. On analysing train data, we found that there are '120'(out of '495') question answering data instances having start_index:0, meaning 120( 25%) question answering data has first word(s) in the context(paragraph) as the answer. We removed 70% of those instances in order to make train data more balanced. In the new train data set we are left with '411' question answering data instances. This time we got the highest test accuracy of 26% at 11 epochs. We have submitted our results for BioASQ test batch-2, got strict accuracy of 32% and our system stood in 2nd place. Initially, hyper-parameter- 'batch size' is set to ‘400’. Later it is tuned to '32'. Although accuracy(exact answer match) remained at 26%, model generated concise and better answers at batch size ‘32’, that is wrong answers are close to the expected answer in good number of cases.
Example.(from BIBREF8)
Question: Which mutated gene causes Chediak Higashi Syndrome?
Exact Answer: ‘lysosomal trafficking regulator gene’.
The answer returned by a model trained at ‘400’ batch size is ‘Autosomal-recessive complicated spastic paraplegia with a novel lysosomal trafficking regulator’, and from the one trained at ‘32’ batch size is ‘lysosomal trafficking regulator’.
In further experiments, we have fine tuned the BioBERT model with both ‘SQuAD’ dataset (version 2.0) and BioAsq train data. For training on ‘SQuAD’, hyper parameters- Learning rate and number of epochs are set to ‘3e-3’ and ‘3’ respectively as mentioned in the paper BIBREF1. Test accuracy of the model boosted to 44%. In one more experiment we trained model only on ‘SQuAD’ dataset, this time test accuracy of the model moved to 47%. The reason model did not perform up to the mark when trained with ‘SQuAD’ alongside BioASQ data could be that in formatted BioASQ data, start_index for the answer is not accurate, and affected the overall accuracy.
Our Systems and Their Performance on Factoid Questions
We have experimented with several systems and their variations, e.g. created by training with specific additional features (see next subsection). Here is their list and short descriptions. Unfortunately we did not pay attention to naming, and the systems evolved between test batches, so the overall picture can only be understood by looking at the details.
When we started the experiments our objective was to see whether BioBERT and entailment-based techniques can provide value for in the context of biomedical question answering. The answer to both questions was a yes, qualified by many examples clearly showing the limitations of both methods. Therefore we tried to address some of these limitations using feature engineering with mixed results: some clear errors got corrected and new errors got introduced, without overall improvement but convincing us that in future experiments it might be worth trying feature engineering again especially if more training data were available.
Overall we experimented with several approaches with the following aspects of the systems changing between batches, that is being absent or present:
training on BioAsq data vs. training on SQuAD
using the BioAsq snippets for context vs. using the documents from the provided URLs for context
adding or not the LAT, i.e. lexical answer type, feature (see BIBREF9, BIBREF10 and an explanation in the subsection just below).
For Yes/No questions (only) we experimented with the entailment methods.
We will discuss the performance of these models below and in Section 6. But before we do that, let us discuss a feature engineering experiment which eventually produced mixed results, but where we feel it is potentially useful in future experiments.
Our Systems and Their Performance on Factoid Questions ::: LAT Feature considered and its impact (slightly negative)
During error analysis we found that for some cases, answer being returned by the model is far away from what it is being asked in the Question.
Example: (from BIBREF8)
Question: Hy's law measures failure of which organ?
Actual Answer: ‘Liver’.
The answer returned by one of our models was ‘alanine aminotransferase’, which is an enzyme. The model returns an enzyme, when the question asked for the organ name. To address this type of errors, we decided to try the concepts of ‘Lexical Answer Type’ (LAT) and Focus Word, which was used in IBM Watson, see BIBREF11 for overview; BIBREF10 for technical details, and BIBREF9 for details on question analysis. In an example given in the last source we read:
POETS & POETRY: He was a bank clerk in the Yukon before he published "Songs of a Sourdough" in 1907.
The focus is the part of the question that is a reference to the answer. In the example above, the focus is "he".
LATs are terms in the question that indicate what type of entity is being asked for.
The headword of the focus is generally a LAT, but questions often contain additional LATs, and in the Jeopardy! domain, categories are an additional source of LATs.
(...) In the example, LATs are "he", "clerk", and "poet".
For example in the question "Which plant does oleuropein originate from?" (BIBREF8). The LAT here is ‘plant’. For the BioAsq task we did not need to explicitly distinguish between the focus and the LAT concepts. In this example, the expectation is that answer returned by the model is a plant. Thus it is conceivable that the cosine distance between contextual embedding of word 'plant' in the question and contextual embedding for the answer present in the paragraph(context) is comparatively low. As a result model learns to adjust its weights during training phase and returns answers with low cosine distance with the LAT.
We used Stanford CoreNLP BIBREF12 library to write rules for extracting lexical answer type present in the question, both 'parts of speech'(POS) and dependency parsing functionality was used. We incorporated the Lexical Answer Type into one of our systems, UNCC_QA1 in Batch 4. This system underperformed our system FACTOIDS by about 3% in the MRR measure, but corrected errors such as in the example above.
Our Systems and Their Performance on Factoid Questions ::: LAT Feature considered and its impact (slightly negative) ::: Assumptions and rules for deriving lexical answer type.
There are different question types: ‘Which’, ‘What’, ‘When’, ‘How’ etc. Each type of question is being handled differently and there are commonalities among the rules written for different question types. Question words are identified through parts of speech tags: 'WDT', 'WRB' ,'WP'. We assumed that LAT is a ‘Noun’ and follows the question word. Often it was also a subject (nsubj). This process is illustrated in Fig.FIGREF15.
LAT computation was governed by a few simple rules, e.g. when a question has multiple words that are 'Subjects’ (and ‘Noun’), a word that is in proximity to the question word is considered as ‘LAT’. These rules are different for each "Wh" word.
Namely, when the word immediately following the question word is a Noun, window size is set to ‘3’. The window size ‘3’ means we iterate through the next ‘3’ words to check if any of the word is both Noun and Subject, If so, such word is considered the ‘LAT’; else the word that is present very next to the question word is considered as the ‘LAT’.
For questions with words ‘Which’ , ‘What’, ‘When’; a Noun immediately following the question word is very often the LAT, e.g. 'enzyme' in Which enzyme is targeted by Evolocumab?. When the word immediately following the question word is not a Noun, e.g. in What is the function of the protein Magt1? the window size is set to ‘5’, and we iterate through the next ‘5’ words (if present) and search for the word that is both Noun and Subject. If present, the word is considered as the ‘LAT’; else, the Noun in close proximity to the question word and following it is returned as the ‘LAT’.
For questions with question words: ‘When’, ‘Who’, ‘Why’, the ’LAT’ is a question word itself. For the word ‘How', e.g. in How many selenoproteins are encoded in the human genome?, we look at the adjective and if we find one, we take it to be the LAT, otherwise the word 'How' is considered as the ‘LAT’.
Perhaps because of using only very simple rules, the accuracy for ‘LAT’ derivation is 75%; that is, in the remaining 25% of the cases the LAT word is identified incorrectly. Worth noting is that the overall performance the system that used LATs was slightly inferior to the system without LATs, but the types of errors changed. The training used BioBERT with the LAT feature as part of the input string. The errors it introduces usually involve finding the wrong element of the correct type e.g. wrong enzyme when two similar enzymes are described in the text, or 'neuron' when asked about a type of cell with a certain function, when the answer calls for a different cell category, adipocytes, and both are mentioned in the text. We feel with more data and additional tuning or perhaps using an ensemble model, we might be able to keep the correct answers, and improve the results on the confusing examples like the one mentioned above. Therefore if we improve our ‘LAT’ derivation logic, or have larger datasets, then perhaps the neural network techniques they will yield better results.
Our Systems and Their Performance on Factoid Questions ::: Impact of Training using BioAsq data (slightly negative)
Training on BioAsq data in our entry in Batch 1 and Batch 2 under the name QA1 showed it might lead to overfitting. This happened both with (Batch 2) and without (Batch 1) hyperparameters tuning: abysmal 18% MRR in Batch 1, and slighly better one, 40% in Batch 2 (although in Batch 2 it was overall the second best result in MRR but 16% lower than the highest score).
In Batch 3 (only), our UNCC_QA3 system was fine tuned on BioAsq and SQuAD 2.0 BIBREF7, and for data preprocessing Context paragraph is generated from relevant snippets provided in the test data. This system underperformed, by about 2% in MRR, our other entry UNCC_QA1, which was also an overall category winner for this batch. The latter was also trained on SQuAD, but not on BioAsq. We suspect that the reason could be the simplistic nature of the find() function described in Section 3.1. So, this could be an area where a better algorithm for finding the best occurrence of an entity could improve performance.
Our Systems and Their Performance on Factoid Questions ::: Impact of Using Context from URLs (negative)
In some experiments, for context in testing, we used documents for which URL pointers are provided in BioAsq. However, our system UNCC_QA3 underperformed our other system tested only on the provided snippets.
In Batch 5 the underperformance was about 6% of MRR, compared to our best system UNCC_QA1, and by 9% to the top performer.
Performance on Yes/No and List questions
Our work focused on Factoid questions. But we also have done experiments on List-type and Yes/No questions.
Performance on Yes/No and List questions ::: Entailment improves Yes/No accuracy
We started by answering always YES (in batch 2 and 3) to get the baseline performance. For batch 4 we used entailment. Our algorithm was very simple: Given a question we iterate through the candidate sentences and try to find any candidate sentence is contradicting the question (with confidence over 50%), if so 'No' is returned as answer, else 'Yes' is returned. In batch 4 this strategy produced better than the BioAsq baseline performance, and compared to our other systems, the use of entailment increased the performance by about 13% (macro F1 score). We used 'AllenNlp' BIBREF13 entailment library to find entailment of the candidate sentences with question.
Performance on Yes/No and List questions ::: For List-type the URLs have negative impact
Overall, we followed the similar strategy that's been followed for Factoid Question Answering task. We started our experiment with batch 2, where we submitted 20 best answers (with context from snippets). Starting with batch 3, we performed post processing: once models generate answer predictions (n-best predictions), we do post-processing on the predicted answers. In test batch 4, our system (called FACTOIDS) achieved highest recall score of ‘0.7033’ but low precision of 0.1119, leaving open the question of how could we have better balanced the two measures.
In the post-processing phase, we take the top ‘20’ (batch 3) and top 5 (batch 4 and 5), predicted answers, tokenize them using common separators: 'comma' , 'and', 'also', 'as well as'. Tokens with characters count more than ‘100’ are eliminated and rest of the tokens are added to the list of possible answers. BioASQ evaluation mechanism does not consider snippets with more than ‘100’ characters as a valid answer. Considering lengthy snippets in to the list of answers would reduce the mean precision score. As a final step, duplicate snippets in the answer pool are removed. For example, consider these top 3 answers predicted by the system (before post-processing):
{
"text": "dendritic cells",
"probability": 0.7554540733426441,
"start_logit": 8.466046333312988,
"end_logit": 9.536355018615723
},
{
"text": "neutrophils, macrophages and
distinct subtypes of dendritic cells",
"probability": 0.13806867348304214,
"start_logit": 6.766478538513184,
"end_logit": 9.536355018615723
},
{
"text": "macrophages and distinct subtypes of dendritic",
"probability": 0.013973475271178242,
"start_logit": 6.766478538513184,
"end_logit": 7.24576473236084
},
After execution of post-processing heuristics, the list of answers returned is as follows:
["dendritic cells"],
["neutrophils"],
["macrophages"],
["distinct subtypes of dendritic cells"]
Summary of our results
The tables below summarize all our results. They show that the performance of our systems was mixed. The simple architectures and algorithm we used worked very well only in Batch 3. However, we feel we can built a better system based on this experience. In particular we observed both the value of contextual embeddings and of feature engineering (LAT), however we failed to combine them properly.
Summary of our results ::: Factoid questions ::: Systems used in Batch 5 experiments
System description for ‘UNCC_QA1’: The system was finetuned on the SQuAD 2.0. For data preprocessing Context / paragraph was generated from relevant snippets provided in the test data.
System description for ‘QA1’ : ‘LAT’ feature was added and finetuned with SQuAD 2.0. For data preprocessing Context / paragraph was generated from relevant snippets provided in the test data.
System description for ‘UNCC_QA3’ : Fine tuning process is same as it is done for the system ‘UNCC_QA1’ in test batch-5. Difference is during data preprocessing, Context/paragraph is generated from the relevant documents for which URLS are included in the test data.
Summary of our results ::: List Questions
For List-type questions, although post processing helped in the later batches, we never managed to obtain competitive precision, although our recall was good.
Summary of our results ::: Yes/No questions
The only thing worth remembering from our performance is that using entailment can have a measurable impact (at least with respect to a weak baseline). The results (weak) are in Table 3.
Discussion, Future Experiments, and Conclusions ::: Summary:
In contrast to 2018, when we submitted BIBREF2 to BioASQ a system based on extractive summarization (and scored very high in the ideal answer category), this year we mainly targeted factoid question answering task and focused on experimenting with BioBERT. After these experiments we see the promise of BioBERT in QA tasks, but we also see its limitations. The latter we tried to address with mixed results using feature engineering. Overall these experiments allowed us to secure a best and a second best score in different test batches. Along with Factoid-type question, we also tried ‘Yes/No’ and ‘List’-type questions, and did reasonably well with our very simple approach.
For Yes/No the moral worth remembering is that reasoning has a potential to influence results, as evidenced by our adding the AllenNLP entailment BIBREF13 system increased its performance.
All our data and software is available at Github, in the previously referenced URL (end of Section 2).
Discussion, Future Experiments, and Conclusions ::: Future experiments
In the current model, we have a shallow neural network with a softmax layer for predicting answer span. Shallow networks however are not good at generalizations. In our future experiments we would like to create dense question answering neural network with a softmax layer for predicting answer span. The main idea is to get contextual word embedding for the words present in the question and paragraph (Context) and feed the contextual word embeddings retrieved from the last layer of BioBERT to the dense question answering network. The mentioned dense layered question answering neural network need to be tuned for finding right hyper parameters. An example of such architecture is shown in Fig.FIGREF30.
In one more experiment, we would like to add a better version of ‘LAT’ contextual word embedding as a feature, along with the actual contextual word embeddings for question text, and Context and feed them as input to the dense question answering neural network. By this experiment, we would like to find if ‘LAT’ feature is improving overall answer prediction accuracy. Adding ‘LAT’ feature this way instead of feeding this word piece embedding directly to the BioBERT (as we did in our above experiments) would not downgrade the quality of contextual word embeddings generated form ‘BioBERT'. Quality contextual word embeddings would lead to efficient transfer learning and chances are that it would improve the model's answer prediction accuracy.
We also see potential for incorporating domain specific inference into the task e.g. using the MedNLI dataset BIBREF14. For all types of experiments it might be worth exploring clinical BERT embeddings BIBREF15, explicitly incorporating domain knowledge (e.g. BIBREF16) and possibly deeper discourse representations (e.g. BIBREF17).
APPENDIX
In this appendix we provide additional details about the implementations.
APPENDIX ::: Systems and their descriptions:
We used several variants of our systems when experimenting with the BioASQ problems. In retrospect, it would be much easier to understand the changes if we adopted some mnemonic conventions in naming the systems. So, we apologize for the names that do not reflect the modifications, and necessitate this list.
APPENDIX ::: Systems and their descriptions: ::: Factoid Type Question Answering:
We preprocessed the test data to convert test data to BioBERT format, We generated Context/paragraph by either aggregating relevant snippets provided or by aggregating documents for which URLS are provided in the BioASQ test data.
APPENDIX ::: Systems and their descriptions: ::: System description for QA1:
We generated Context/paragraph by aggregating relevant snippets available in the test data and mapped it against the question text and question id. We ignored the content present in the documents (document URLS were provided in the original test data). The model is finetuned with BioASQ data.
data preprocessing is done in the same way as it is done for test batch-1. Model fine tuned on BioASQ data.
‘LAT’/ Focus word feature added and fine tuned with SQuAD 2.0 [reference]. For data preprocessing Context / paragraph is generated from relevant snippets provided in the test data.
APPENDIX ::: Systems and their descriptions: ::: System description for UNCC_QA_1:
System is finetuned on the SQuAD 2.0 [reference]. For data preprocessing Context / paragraph is generated from relevant snippets provided in the test data.
‘LAT’/ Focus word feature added and fine tuned with SQuAD 2.0 [reference]. For data preprocessing Context / paragraph is generated from relevant snippets provided in the test data.
The System is finetuned on the SQuAD 2.0. For data preprocessing Context / paragraph is generated from relevant snippets provided in the test data.
APPENDIX ::: Systems and their descriptions: ::: System description for UNCC_QA3:
System is finetuned on the SQuAD 2.0 [reference] and BioASQ dataset[].For data preprocessing Context / paragraph is generated from relevant snippets provided in the test data.
Fine tuning process is same as it is done for the system ‘UNCC_QA_1’ in test batch-5. Difference is during data preprocessing, Context/paragraph is generated form from the relevant documents for which URLS are included in the test data.
APPENDIX ::: Systems and their descriptions: ::: System description for UNCC_QA2:
Fine tuning process is same as for ‘UNCC_QA_1 ’. Difference is Context/paragraph is generated form from the relevant documents for which URLS are included in the test data. System ‘UNCC_QA_1’ got the highest ‘MRR’ score in the 3rd test batch set.
APPENDIX ::: Systems and their descriptions: ::: System description for FACTOIDS:
The System is finetuned on the SQuAD 2.0. For data preprocessing Context / paragraph is generated from relevant snippets provided in the test data.
APPENDIX ::: Systems and their descriptions: ::: List Type Questions:
We attempted List type questions starting from test batch ‘2’. Used similar approach that's been followed for Factoid Question answering task. For all the test batch sets, in the data pre processing phase Context/ paragraph is generated either by aggregating relevant snippets or by aggregating documents(URLS) provided in the BioASQ test data.
For test batch-2, model (System: QA1) is finetuned on BioASQ data and submitted top ‘20’ answers predicted by the model as the list of answers. system ‘QA1’ achieved low F-Measure score:‘0.0786’ in the second test batch. In the further test batches for List type questions, we finetuned the model on Squad data set [reference], implemented post processing techniques (refer section 5.2) and achieved a better F-measure score: ‘0.2862’ in the final test batch set.
In test batch-3 (Systems : ‘QA1’/’’UNCC_QA_1’/’UNCC_QA3’/’UNCC_QA2’) top 20 answers returned by the model is sent for post processing and in test batch 4 and 5 only top 5 answers are sent for post processing. System UNCC_QA2(in batch 3) for List type question answering, Context is generated from documents for which URLS are provided in the BioASQ test data. for the rest of the systems (in test batch-3) for List Type question answering task snippets present in the BioaSQ test data are used to generate context.
In test batch-4 (System : ‘FACTOIDS’/’UNCC_QA_1’/’UNCC_QA3’’) top 5 answers returned by the model is sent for post processing. In case of system ‘FACTOIDS’ snippets in the test data were used to generate context. for systems ’UNCC_QA_1’ and ’UNCC_QA3’ context is generated from the documents for which URLS are provided in the BioASQ test data.
In test batch-5 ( Systems: ‘QA1’/’UNCC_QA_1’/’UNCC_QA3’/’UNCC_QA2’ ) our approach is the same as that of test batch-4 where top 5 answers returned by the model is sent for post processing. for all the systems (in test batch-5) context is generated from the snippets provided in the BioASQ test data.
APPENDIX ::: Systems and their descriptions: ::: Yes/No Type Questions:
For the first 3 test batches, We have submitted answer ‘Yes’ to all the questions. Later, we employed ‘Sentence Entailment’ techniques(refer section 6.0) for the fourth and fifth test batch sets. Our Systems with ‘Sentence Entailment’ approach (for ‘Yes’/ ‘No’ question answering): ‘UNCC_QA_1’(test batch-4), UNCC_QA3(test batch-5).
APPENDIX ::: Additional details for Yes/No Type Questions
We used Textual Entailment in Batch 4 and 5 for ‘Yes’/‘No’ question type. The algorithm was very simple: Given a question we iterate through the candidate sentences, and look for any candidate sentences contradicting the question. If we find one 'No' is returned as answer, else 'Yes' is returned. (The confidence for contradiction was set at 50%) We used AllenNLP BIBREF13 entailment library to find entailment of the candidate sentences with question.
Flow Chart for Yes/No Question answer processing is shown in Fig.FIGREF51
APPENDIX ::: Assumptions, rules and logic flow for deriving Lexical Answer Types from questions
There are different question types, and we distinguished them based on the question words: ‘Which’, ‘What’, ‘When’, ‘How’ etc. Each type of question is being handled differently and there are commonalities among the rules written for different question types. How are question words identified? question words have parts of speech(POS): 'WDT', 'WRB', 'WP'.
Assumptions:
1) Lexical answer type (‘LAT’) or focus word is of type Noun and follows the question word.
2) The LAT word is a Subject. (This clearly not always true, but we used a very simple method). Note: ‘StanfordNLP’ dependency parsing tag for identifying subject is 'nsubj' or 'nsubjpass'.
3) When a question has multiple words that are of type Subject (and Noun), a word that is in proximity to the question word is considered as ‘LAT’.
4) For questions with question words: ‘When’, ‘Who’, ‘Why’, the ’LAT’ is a question word itself that is, ‘When’, ‘Who’, ‘Why’ respectively.
Rules and logic flow to traverse a question: The three cases below describe the logic flow of finding LATs. The figures show the grammatical structures used for this purpose.
APPENDIX ::: Assumptions, rules and logic flow for deriving Lexical Answer Types from questions ::: Case-1:
Question with question word ‘How’.
For questions with question word 'How', the adjective that follows the question word is considered as ‘LAT’ (need not follow immediately). If an adjective is absent, word 'How' is considered as ‘LAT’. When there are multiple words that are adjectives, a word in close proximity to the question word and follows it is returned as ‘LAT’. Note: The part of speech tag to identify adjectives is 'JJ'. For Other possible question words like ‘whose’. ‘LAT’/Focus word is question words itself.
Example Question: How many selenoproteins are encoded in the human genome?
APPENDIX ::: Assumptions, rules and logic flow for deriving Lexical Answer Types from questions ::: Case-2:
Questions with question words ‘Which’ , ‘What’ and all other possible question words; a 'Noun' immediately following the question word.
Example Question: Which enzyme is targeted by Evolocumab?
Here, Focus word/LAT is ‘enzyme’ which is both Noun and Subject and immediately follows the question word.
When the word immediately following the question word is a noun, the window size is set to ‘3’. This size ‘3’ means that we iterate through the next ‘3’ words (if present) to check if any of the word is both 'Noun' and 'Subject', If so, the word is considered as ‘LAT’/Focus Word. Else the word that is present very next to the question word is considered as ‘LAT’.
APPENDIX ::: Assumptions, rules and logic flow for deriving Lexical Answer Types from questions ::: Case-3:
Questions with question words ‘Which’ , ‘What’ and all other possible question words; word immediately following the question word is not a 'Noun'.
Example Question: What is the function of the protein Magt1?
Here, Focus word/LAT is ‘function ’ which is both Noun and Subject and does not immediately follow the question word.
When the very next word following the question word is not a Noun, window size is set to ‘5’. Window size ‘5’ corresponds that we iterate through the next ‘5’ words (if present) and search for the word that is both Noun and Subject. If present, the word is considered as ‘LAT’. Else, the 'Noun' close proximity to the question word and follows it is returned as ‘LAT’.
Ad we mentioned earlier, the accuracy for ‘LAT’ derivation is 75 percent. But clearly the simple logic described above can be improved, as shown in BIBREF9, BIBREF10. Whether this in turn produces improvements in this particular task is an open question.
APPENDIX ::: Proposing Future Experiments
In the current model, we have a shallow neural network with a softmax layer for predicting answer span. Shallow networks however are not good at generalizations. In our future experiments we would like to create dense question answering neural network with a softmax layer for predicting answer span. The main idea is to get contextual word embedding for the words present in the question and paragraph (Context) and feed the contextual word embeddings retrieved from the last layer of BioBERT to the dense question answering network. The mentioned dense layered question answering Neural network need to be tuned for finding right hyper parameters. An example of such architecture is shown in Fig.FIGREF30.
In another experiment we would like to only feed contextual word embeddings for Focus word/ ‘LAT’, paragraph/ Context as input to the question answering neural network. In this experiment we would neglect all embeddings for the question text except that of Focus word/ ‘LAT’. Our assumption and idea for considering focus word and neglecting remaining words in the question is that during training phase it would make more precise for the model to identify the focus of the question and map answers against the question’s focus. To validate our assumption, we would like to take sample question answering data and find the cosine distance between contextual embedding of Focus word and that of the actual answer and verify if the cosine distance is comparatively low in most of the cases.
In one more experiment, we would like to add a better version of ‘LAT’ contextual word embedding as a feature, along with the actual contextual word embeddings for question text, and Context and feed them as input to the dense question answering neural network. By this experiment, we would like to find if ‘LAT’ feature is improving overall answer prediction accuracy. Adding ‘LAT’ feature this way instead of feeding Focus word’s word piece embedding directly (as we did in our above experiments) to the BioBERT would not downgrade the quality of contextual word embeddings generated form ‘BioBERT'. Quality contextual word embeddings would lead to efficient transfer learning and chances are that it would improve the model's answer prediction accuracy.
Question:
What was the baseline model?
Answer:
| by answering always YES (in batch 2 and 3) | 328 | 43,312 | 43,314 | 43,352 | ... [The rest of the article is omitted]
|
1b1a30e9e68a9ae76af467e60cefb180d135e285 | 1b1a30e9e68a9ae76af467e60cefb180d135e285_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Problem Statement
Spoken conversations still remain the most natural and effortless means of human communication. Thus a lot of valuable information is conveyed and exchanged in such an unstructured form. In telehealth settings, nurses might call discharged patients who have returned home to continue to monitor their health status. Human language technology that can efficiently and effectively extract key information from such conversations is clinically useful, as it can help streamline workflow processes and digitally document patient medical information to increase staff productivity. In this work, we design and prototype a dialogue comprehension system in the question-answering manner, which is able to comprehend spoken conversations between nurses and patients to extract clinical information.
Motivation of Approach
Machine comprehension of written passages has made tremendous progress recently. Large quantities of supervised training data for reading comprehension (e.g. SQuAD BIBREF0 ), the wide adoption and intense experimentation of neural modeling BIBREF1 , BIBREF2 , and the advancements in vector representations of word embeddings BIBREF3 , BIBREF4 all contribute significantly to the achievements obtained so far. The first factor, the availability of large scale datasets, empowers the latter two factors. To date, there is still very limited well-annotated large-scale data suitable for modeling human-human spoken dialogues. Therefore, it is not straightforward to directly port over the recent endeavors in reading comprehension to dialogue comprehension tasks.
In healthcare, conversation data is even scarcer due to privacy issues. Crowd-sourcing is an efficient way to annotate large quantities of data, but less suitable for healthcare scenarios, where domain knowledge is required to guarantee data quality. To demonstrate the feasibility of a dialogue comprehension system used for extracting key clinical information from symptom monitoring conversations, we developed a framework to construct a simulated human-human dialogue dataset to bootstrap such a prototype. Similar efforts have been conducted for human-machine dialogues for restaurant or movie reservations BIBREF5 . To the best of our knowledge, no one to date has done so for human-human conversations in healthcare.
Human-human Spoken Conversations
Human-human spoken conversations are a dynamic and interactive flow of information exchange. While developing technology to comprehend such spoken conversations presents similar technical challenges as machine comprehension of written passages BIBREF6 , the challenges are further complicated by the interactive nature of human-human spoken conversations:
(1) Zero anaphora is more common: Co-reference resolution of spoken utterances from multiple speakers is needed. For example, in Figure FIGREF5 (a) headaches, the pain, it, head bulging all refer to the patient's headache symptom, but they were uttered by different speakers and across multiple utterances and turns. In addition, anaphors are more likely to be omitted (see Figure FIGREF5 (a) A4) as this does not affect the human listener’s understanding, but it might be challenging for computational models.
(2) Thinking aloud more commonly occurs: Since it is more effortless to speak than to type, one is more likely to reveal her running thoughts when talking. In addition, one cannot retract what has been uttered, while in text communications, one is more likely to confirm the accuracy of the information in a written response and revise if necessary before sending it out. Thinking aloud can lead to self-contradiction, requiring more context to fully understand the dialogue; e.g., in A6 in Figure FIGREF5 (a), the patient at first says he has none of the symptoms asked, but later revises his response saying that he does get dizzy after running.
(3) Topic drift is more common and harder to detect in spoken conversations: An example is shown in Figure FIGREF5 (a) in A3, where No is actually referring to cough in the previous question, and then the topic is shifted to headache. In spoken conversations, utterances are often incomplete sentences so traditional linguistic features used in written passages such as punctuation marks indicating syntactic boundaries or conjunction words suggesting discourse relations might no longer exist.
Dialogue Comprehension Task
Figure FIGREF5 (b) illustrates the proposed dialogue comprehension task using a question answering (QA) model. The input are a multi-turn symptom checking dialogue INLINEFORM0 and a query INLINEFORM1 specifying a symptom with one of its attributes; the output is the extracted answer INLINEFORM2 from the given dialogue. A training or test sample is defined as INLINEFORM3 . Five attributes, specifying certain details of clinical significance, are defined to characterize the answer types of INLINEFORM4 : (1) the time the patient has been experiencing the symptom, (2) activities that trigger the symptom (to occur or worsen), (3) the extent of seriousness, (4) the frequency occurrence of the symptom, and (5) the location of symptom. For each symptom/attribute, it can take on different linguistic expressions, defined as entities. Note that if the queried symptom or attribute is not mentioned in the dialogue, the groundtruth output is “No Answer”, as in BIBREF6 .
Reading Comprehension
Large-scale reading comprehension tasks like SQuAD BIBREF0 and MARCO BIBREF7 provide question-answer pairs from a vast range of written passages, covering different kinds of factual answers involving entities such as location and numerical values. Furthermore, HotpotQA BIBREF8 requires multi-step inference and provides numerous answer types. CoQA BIBREF9 and QuAC BIBREF10 are designed to mimic multi-turn information-seeking discussions of the given material. In these tasks, contextual reasoning like coreference resolution is necessary to grasp rich linguistic patterns, encouraging semantic modeling beyond naive lexical matching. Neural networks contribute to impressive progress in semantic modeling: distributional semantic word embeddings BIBREF3 , contextual sequence encoding BIBREF11 , BIBREF12 and the attention mechanism BIBREF13 , BIBREF14 are widely adopted in state-of-the-art comprehension models BIBREF1 , BIBREF2 , BIBREF4 .
While language understanding tasks in dialogue such as domain identification BIBREF15 , slot filling BIBREF16 and user intent detection BIBREF17 have attracted much research interest, work in dialogue comprehension is still limited, if any. It is labor-intensive and time-consuming to obtain a critical mass of annotated conversation data for computational modeling. Some propose to collect text data from human-machine or machine-machine dialogues BIBREF18 , BIBREF5 . In such cases, as human speakers are aware of current limitations of dialogue systems or due to pre-defined assumptions of user simulators, there are fewer cases of zero anaphora, thinking aloud, and topic drift, which occur more often in human-human spoken interactions.
NLP for Healthcare
There is emerging interest in research and development activities at the intersection of machine learning and healthcare , of which much of the NLP related work are centered around social media or online forums (e.g., BIBREF19 , BIBREF20 ), partially due to the world wide web as a readily available source of information. Other work in this area uses public data sources such as MIMIC in electronic health records: text classification approaches have been applied to analyze unstructured clinical notes for ICD code assignment BIBREF21 and automatic intensive emergency prediction BIBREF22 . Sequence-to-sequence textual generation has been used for readable notes based on medical and demographic recordings BIBREF23 . For mental health, there has been more focus on analyzing dialogues. For example, sequential modeling of audio and text have helped detect depression from human-machine interviews BIBREF24 . However, few studies have examined human-human spoken conversations in healthcare settings.
Data Preparation
We used recordings of nurse-initiated telephone conversations for congestive heart failure patients undergoing telemonitoring, post-discharge from the hospital. The clinical data was acquired by the Health Management Unit at Changi General Hospital. This research study was approved by the SingHealth Centralised Institutional Review Board (Protocol 1556561515). The patients were recruited during 2014-2016 as part of their routine care delivery, and enrolled into the telemonitoring health management program with consent for use of anonymized versions of their data for research.
The dataset comprises a total of 353 conversations from 40 speakers (11 nurses, 16 patients, and 13 caregivers) with consent to the use of anonymized data for research. The speakers are 38 to 88 years old, equally distributed across gender, and comprise a range of ethnic groups (55% Chinese, 17% Malay, 14% Indian, 3% Eurasian, and 11% unspecified). The conversations cover 11 topics (e.g., medication compliance, symptom checking, education, greeting) and 9 symptoms (e.g., chest pain, cough) and amount to 41 hours.
Data preprocessing and anonymization were performed by a data preparation team, separate from the data analysis team to maintain data confidentiality. The data preparation team followed standard speech recognition transcription guidelines, where words are transcribed verbatim to include false starts, disfluencies, mispronunciations, and private self-talk. Confidential information were marked and clipped off from the audio and transcribed with predefined tags in the annotation. Conversation topics and clinical symptoms were also annotated and clinically validated by certified telehealth nurses.
Linguistic Characterization on Seed Data
To analyze the linguistic structure of the inquiry-response pairs in the entire 41-hour dataset, we randomly sampled a seed dataset consisting of 1,200 turns and manually categorized them to different types, which are summarized in Table TABREF14 along with the corresponding occurrence frequency statistics. Note that each given utterance could be categorized to more than one type. We elaborate on each utterance type below.
Open-ended Inquiry: Inquiries about general well-being or a particular symptom; e.g., “How are you feeling?” and “Do you cough?”
Detailed Inquiry: Inquiries with specific details that prompt yes/no answers or clarifications; e.g., “Do you cough at night?”
Multi-Intent Inquiry: Inquiring more than one symptom in a question; e.g., “Any cough, chest pain, or headache?”
Reconfirmation Inquiry: The nurse reconfirms particular details; e.g., “Really? At night?” and “Serious or mild?”. This case is usually related to explicit or implicit coreferencing.
Inquiry with Transitional Clauses: During spoken conversations, one might repeat what the other party said, but it is unrelated to the main clause of the question. This is usually due to private self-talk while thinking aloud, and such utterances form a transitional clause before the speaker starts a new topic; e.g., “Chest pain... no chest pain, I see... any cough?”.
Yes/No Response: Yes/No responses seem straightforward, but sometimes lead to misunderstanding if one does not interpret the context appropriately. One case is tag questions: A:“You don't cough at night, do you?” B:`Yes, yes” A:“cough at night?” B:“No, no cough”. Usually when the answer is unclear, clarifying inquiries will be asked for reconfirmation purposes.
Detailed Response: Responses that contain specific information of one symptom, like “I felt tightness in my chest”.
Response with Revision: Revision is infrequent but can affect comprehension significantly. One cause is thinking aloud so a later response overrules the previous one; e.g., “No dizziness, oh wait... last week I felt a bit dizzy when biking”.
Response with Topic Drift: When a symptom/topic like headache is inquired, the response might be: “Only some chest pain at night”, not referring to the original symptom (headache) at all.
Response with Transitional Clauses: Repeating some of the previous content, but often unrelated to critical clinical information and usually followed by topic drift. For example, “Swelling... swelling... I don't cough at night”.
Simulating Symptom Monitoring Dataset for Training
We divide the construction of data simulation into two stages. In Section SECREF16 , we build templates and expression pools using linguistic analysis followed by manual verification. In Section SECREF20 , we present our proposed framework for generating simulated training data. The templates and framework are verified for logical correctness and clinical soundness.
Template Construction
Each utterance in the seed data is categorized according to Table TABREF14 and then abstracted into templates by replacing entity phrases like cough and often with respective placeholders “#symptom#” and “#frequency#”. The templates are refined through verifying logical correctness and injecting expression diversity by linguistically trained researchers. As these replacements do not alter the syntactic structure, we interchange such placeholders with various verbal expressions to enlarge the simulated training set in Section SECREF20 . Clinical validation was also conducted by certified telehealth nurses.
For the 9 symptoms (e.g. chest pain, cough) and 5 attributes (e.g., extent, frequency), we collect various expressions from the seed data, and expand them through synonym replacement. Some attributes are unique to a particular symptom; e.g., “left leg” in #location# is only suitable to describe the symptom swelling, but not the symptom headache. Therefore, we only reuse general expressions like “slight” in #extent# across different symptoms to diversify linguistic expressions.
Two linguistically trained researchers constructed expression pools for each symptom and each attribute to account for different types of paraphrasing and descriptions. These expression pools are used in Section SECREF20 (c).
Simulated Data Generation Framework
Figure FIGREF15 shows the five steps we use to generate multi-turn symptom monitoring dialogue samples.
(a) Topic Selection: While nurses might prefer to inquire the symptoms in different orders depending on the patient's history, our preliminary analysis shows that modeling results do not differ noticeably if topics are of equal prior probabilities. Thus we adopt this assumption for simplicity.
(b) Template Selection: For each selected topic, one inquiry template and one response template are randomly chosen to compose a turn. To minimize adverse effects of underfitting, we redistributed the frequency distribution in Table TABREF14 : For utterance types that are below 15%, we boosted them to 15%, and the overall relative distribution ranking is balanced and consistent with Table TABREF14 .
(c) Enriching Linguistic Expressions: The placeholders in the selected templates are substituted with diverse expressions from the expression pools in Section UID19 to characterize the symptoms and their corresponding attributes.
(d) Multi-Turn Dialogue State Tracking: A greedy algorithm is applied to complete conversations. A “completed symptoms” list and a “to-do symptoms” list are used for symptom topic tracking. We also track the “completed attributes" and “to-do attributes". For each symptom, all related attributes are iterated. A dialogue ends only when all possible entities are exhausted, generating a multi-turn dialogue sample, which encourages the model to learn from the entire discussion flow rather than a single turn to comprehend contextual dependency. The average length of a simulated dialogue is 184 words, which happens to be twice as long as an average dialogue from the real-world evaluation set. Moreover, to model the roles of the respondents, we set the ratio between patients and caregivers to 2:1; this statistic is inspired by the real scenarios in the seed dataset. For both the caregivers and patients, we assume equal probability of both genders. The corresponding pronouns in the conversations are thus determined by the role and gender of these settings.
(e) Multi-Turn Sample Annotation: For each multi-turn dialogue, a query is specified by a symptom and an attribute. The groundtruth output of the QA system is automatically labeled based on the template generation rules, but also manually verified to ensure annotation quality. Moreover, we adopt the unanswerable design in BIBREF6 : when the patient does not mention a particular symptom, the answer is defined as “No Answer”. This process is repeated until all logical permutations of symptoms and attributes are exhausted.
Experiments
Model Design
We implemented an established model in reading comprehension, a bi-directional attention pointer network BIBREF1 , and equipped it with an answerable classifier, as depicted in Figure FIGREF21 . First, tokens in the given dialogue INLINEFORM0 and query INLINEFORM1 are converted into embedding vectors. Then the dialogue embeddings are fed to a bi-directional LSTM encoding layer, generating a sequence of contextual hidden states. Next, the hidden states and query embeddings are processed by a bi-directional attention layer, fusing attention information in both context-to-query and query-to-context directions. The following two bi-directional LSTM modeling layers read the contextual sequence with attention. Finally, two respective linear layers with softmax functions are used to estimate token INLINEFORM2 's INLINEFORM3 and INLINEFORM4 probability of the answer span INLINEFORM5 .
In addition, we add a special tag “[SEQ]” at the head of INLINEFORM0 to account for the case of “No answer” BIBREF4 and adopt an answerable classifier as in BIBREF25 . More specifically, when the queried symptom or attribute is not mentioned in the dialogue, the answer span should point to the tag “[SEQ]” and answerable probability should be predicted as 0.
Implementation Details
The model was trained via gradient backpropagation with the cross-entropy loss function of answer span prediction and answerable classification, optimized by Adam algorithm BIBREF26 with initial learning rate of INLINEFORM0 . Pre-trained GloVe BIBREF3 embeddings (size INLINEFORM1 ) were used. We re-shuffled training samples at each epoch (batch size INLINEFORM2 ). Out-of-vocabulary words ( INLINEFORM3 ) were replaced with a fixed random vector. L2 regularization and dropout (rate INLINEFORM4 ) were used to alleviate overfitting BIBREF27 .
Evaluation Setup
To evaluate the effectiveness of our linguistically-inspired simulation approach, the model is trained on the simulated data (see Section SECREF20 ). We designed 3 evaluation sets: (1) Base Set (1,264 samples) held out from the simulated data. (2) Augmented Set (1,280 samples) built by adding two out-of-distribution symptoms, with corresponding dialogue contents and queries, to the Base Set (“bleeding” and “cold”, which never appeared in training data). (3) Real-World Set (944 samples) manually delineated from the the symptom checking portions (approximately 4 hours) of real-world dialogues, and annotated as evaluation samples.
Results
Evaluation results are in Table TABREF25 with exact match (EM) and F1 score in BIBREF0 metrics. To distinguish the correct answer span from the plausible ones which contain the same words, we measure the scores on the position indices of tokens. Our results show that both EM and F1 score increase with training sample size growing and the optimal size in our setting is 100k. The best-trained model performs well on both the Base Set and the Augmented Set, indicating that out-of-distribution symptoms do not affect the comprehension of existing symptoms and outputs reasonable answers for both in- and out-of-distribution symptoms. On the Real-World Set, we obtained 78.23 EM score and 80.18 F1 score respectively.
Error analysis suggests the performance drop from the simulated test sets is due to the following: 1) sparsity issues resulting from the expression pools excluding various valid but sporadic expressions. 2) nurses and patients occasionally chit-chat in the Real-World Set, which is not simulated in the training set. At times, these chit-chats make the conversations overly lengthy, causing the information density to be lower. These issues could potentially distract and confuse the comprehension model. 3) an interesting type of infrequent error source, caused by patients elaborating on possible causal relations of two symptoms. For example, a patient might say “My giddiness may be due to all this cough”. We are currently investigating how to close this performance gap efficiently.
Ablation Analysis
To assess the effectiveness of bi-directional attention, we bypassed the bi-attention layer by directly feeding the contextual hidden states and query embeddings to the modeling layer. To evaluate the pre-trained GloVe embeddings, we randomly initialized and trained the embeddings from scratch. These two procedures lead to 10% and 18% performance degradation on the Augmented Set and Real-World Set, respectively (see Table TABREF27 ).
Conclusion
We formulated a dialogue comprehension task motivated by the need in telehealth settings to extract key clinical information from spoken conversations between nurses and patients. We analyzed linguistic characteristics of real-world human-human symptom checking dialogues, constructed a simulated dataset based on linguistically inspired and clinically validated templates, and prototyped a QA system. The model works effectively on a simulated test set using symptoms excluded during training and on real-world conversations between nurses and patients. We are currently improving the model's dialogue comprehension capability in complex reasoning and context understanding and also applying the QA model to summarization and virtual nurse applications.
Acknowledgements
Research efforts were supported by funding for Digital Health and Deep Learning I2R (DL2 SSF Project No: A1718g0045) and the Science and Engineering Research Council (SERC Project No: A1818g0044), A*STAR, Singapore. In addition, this work was conducted using resources and infrastructure provided by the Human Language Technology unit at I2R. The telehealth data acquisition was funded by the Economic Development Board (EDB), Singapore Living Lab Fund and Philips Electronics – Hospital to Home Pilot Project (EDB grant reference number: S14-1035-RF-LLF H and W).
We acknowledge valuable support and assistance from Yulong Lin, Yuanxin Xiang, and Ai Ti Aw at the Institute for Infocomm Research (I2R); Weiliang Huang at the Changi General Hospital (CGH) Department of Cardiology, Hong Choon Oh at CGH Health Services Research, and Edris Atikah Bte Ahmad, Chiu Yan Ang, and Mei Foon Yap of the CGH Integrated Care Department.
We also thank Eduard Hovy and Bonnie Webber for insightful discussions and the anonymous reviewers for their precious feedback to help improve and extend this piece of work.
Question:
How big is their created dataset?
Answer:
| 353 conversations from 40 speakers (11 nurses, 16 patients, and 13 caregivers), we build templates and expression pools using linguistic analysis | 328 | 23,440 | 23,442 | 23,485 | ... [The rest of the article is omitted]
|
dea9e7fe8e47da5e7f31d9b1a46ebe34e731a596 | dea9e7fe8e47da5e7f31d9b1a46ebe34e731a596_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
“Laughter is the best Medicine” is a saying which is popular with most of the people. Humor is a form of communication that bridges the gap between various languages, cultures, ages and demographics. That's why humorous content with funny and witty hashtags are so much popular on social media. It is a very powerful tool to connect with the audience. Automatic Humor Recognition is the task of determining whether a text contains some level of humorous content or not. First conference on Computational humor was organized in 1996, since then many research have been done in this field. kao2016computational does pun detection in one-liners and dehumor detects humor in Yelp reviews. Because of the complex and interesting aspects involved in detecting humor in texts, it is one of the challenging research field in Natural Language Processing BIBREF3 . Identifying humor in a sentence sometimes require a great amount of external knowledge to completely understand it. There are many types of humor, namely anecdotes, fantasy, insult, irony, jokes, quote, self deprecation etc BIBREF4 , BIBREF5 . Most of the times there are different meanings hidden inside a sentence which is grasped differently by individuals, making the task of humor identification difficult, which is why the development of a generalized algorithm to classify different type of humor is a challenging task.
Majority of the researches on social media texts is focused on English. A study by schroeder2010half shows that, a high percentage of these texts are in non-English languages. fischer2011language gives some interesting information about the languages used on Twitter based on the geographical locations. With a huge amount of such user generated data available on social media, there is a need to develop technologies for non-English languages. In multilingual regions like South Asia, majority of the social media users speak more than two languages. In India, Hindi is the most spoken language (spoken by 41% of the population) and English is the official language of the country. Twitter has around 23.2 million monthly active users in India. Native speakers of Hindi often put English words in the sentences and transliterate the whole sentence to Latin script while posting on social media, thereby making the task of automatic text classification a very challenging problem. Linguists came up with a term for any type of language mixing, known as `code-mixing' or `code-switching' BIBREF6 , BIBREF7 , BIBREF8 , BIBREF9 . Both the terms are used interchangeably, but there is a slight difference between the two terms. Code-mixing refers to the insertion of words, phrases, and morphemes of one language into a statement or an expression of another language, whereas transliteration of every word in a sentence to another script ( here Devanagari to Latin) is coined code-switching BIBREF10 . The first tweet in Figure 1 is an example of code-mixing and second is an example of code-switching. In this paper, we use code-mixing to denote both cases.
In this paper, we present a freely available corpus containing code-mixed tweets in Hindi and English language with tweets written in Latin script. Tweets are manually classified into humorous and non-humorous classes. Moreover, each token in the tweets is also given a language tag which determines the source or origin language of the token (English or Hindi). The paper is divided in sections as follows, we start by describing the corpus and the annotation scheme in Section 2. Section 3 summarizes our supervised classification system which includes pre-processing of the tweets in the dataset and the feature extraction followed by the method used to identify humor in tweets. In the next subsection, we describe the classification model and the results of the experiments conducted using character and word level features. In the last section, we conclude the paper followed by future work and references.
Corpus Creation and Annotation
In this section, we explain the techniques used in the creation and annotation of the corpus.
Data Collection
Python package twitterscraper is used to scrap tweets from twitter. 10,478 tweets from the past two years from domains like `sports', `politics', `entertainment' were extracted. Among those tweets, we manually removed the tweets which were written either in English or Hindi entirely. There were 4161 tweets written in English and 2774 written in Hindi. Finally, a total of 3543 English-Hindi code-mixed tweets were collected. Table 1 describes the number of tweets and words in each category.
We tried to make the corpus balanced i.e. uniform distribution of tweets in each category to yield better supervised classification results as described by BIBREF11 .
Humor Annotation
The final code-mixed tweets were forwarded to a group of three annotators who were university students and fluent in both English and Hindi. Approximately 60 hours were spent in tagging tweets for the presence of humor. Tweets which consisted of any anecdotes, fantasy, irony, jokes, insults were annotated as humorous whereas tweets stating any facts, dialogues or speech which did not contain amusement were put in non-humorous class. Following are some examples of code-mixed tweets in the corpus:
“For #WontGiveItBack to work, Dhoni needs to say `Trophy toh ghar par hi bhul aaye' ”
(For #WontGiveItBack to work, Dhoni needs to say “We forgot the trophy at home”) .
Na sadak na naukri bas badh rahi gundagardi #FailedCMNitish
(No roads no naukri, only hooliganism increasing #FailedCMNitish )
Subha ka bhula agar sham ko wapas ghar aa jaye then we must thank GPS technology.
(If someone lost in the morning return home in the evening then we must thank GPS technology.)
She : Gori hai kalaiyan pehna de mujhe hari hari chudiya He *gets green bangles* She : Not this green ya, bottle green color.
(She : (sings a song) My wrists are so beautiful, give me green green bangles He *gets green bangles* She : Not this green ya, bottle green color. )
Dard dilon ke kam ho jaate... twitter par agar poetries kam ho jaate
(pain in the heart will reduce...if number of poetries decreases on twitter).
Annotators were given certain guidelines to decide whether a tweet was humorous or not. The context of the tweet could be found by searching about hashtag or keywords used in the tweet. Example (1) uses a hashtag `#WontGiveItBack' which was trending during the ICC cricket world cup 2015. Searching it on Google gave 435k results and the time of the tweet was after the final match of the tournament. So there is an observational humor in (1) as India won the world cup in 2011 and lost in 2015 , hence the tweet was classified as humorous. Any tweets stating any facts, news or reality were classified as non-humorous. There were many tweets which did not contain any hashtags, to understand the context of such tweets annotators selected some keywords from the tweet and searched them online. Example (2) contains a comment towards a political leader towards development and was categorized as non-humorous. Tweets containing normal jokes and funny quotes like in (3) and (4) were put in humorous category. There were some tweets like (5) which consists of poem or lines of a song but modified. Annotators were guided that if such tweets contains satire or any humoristic features, then it could be categorized as humorous otherwise not. There were some tweets which were typical to categorize like (5), hence it was left to the annotators to the best of their understanding. Based on the above guidelines annotators categorized the tweets. To measure inter annotator agreement we opted for Fleiss' Kappa BIBREF12 obtaining an agreement of 0.821 . Both humorous and non-humorous tweets in nearly balanced amount were selected to prepare the corpus. If we had included humorous tweets from one domain like sports and non humorous tweets from another domain like news then, it would have given high performance of classification BIBREF13 . To classify based on the semantics and not on the domain differences, we included both types of tweets from different domains. Many tweets contains a picture along with a caption. Sometimes a caption may not contain humor but combined with the picture, it can provide some degree of humor. Such tweets were removed from the corpus to make the corpus unimodal. In Figure 1, the first tweet, “Anurag Kashyap can never join AAP because ministers took oath `main kisi Anurag aur dwesh ke bina kaam karunga' ” (Anurag Kashyap can never join AAP because ministers took oath `I will work without any affection (Anurag in Hindi) and without hesitation (dwesh in Hindi)'), was classified as humorous. The second tweet, “#SakshiMalik take a bow! #proudIndian #Rio #Olympics #BronzeMedal #girlpower Hamaara khaata khul Gaya!” (#SakshiMalik take a bow! #proudIndian #Rio #Olympics #BronzeMedal #girlpower Our account opened!) was classified as non-humorous as it contains a pride statement.
Language Annotation
Code-mixing provides some challenges for language identification like spelling variations and non-grammatical structures in a sentence BIBREF0 . Therefore, we annotated the tweets with the language at the word level. Native speakers of Hindi and proficient in English, labelled the language of the tokens in the tweets. Three types of tags were assigned to the tokens , En is assigned to the tokens present in English vocabulary like “family”, “Children” etc. Similarly, Hi is assigned to the tokens present in Hindi vocabulary but transliterated to Latin script like “samay” (time), “aamaadmi” (common man). Rest of the tokens consists of proper nouns, numbers, dates, urls, hashtags, mentions, emojis and punctuations which are labelled as Ot(others). Major concern in language annotation was to annotate the words present in both languages (ambiguous words). For example, `to' (`but' in Hindi) and `is' (`this' in Hindi) , for this scenario annotators understood the context of the tweet and based on that the words were being annotated. For example, consider two sentences `This place is 500 years old' and `Is bar Modi Sarkar'(This time Modi Government). So `is' in first sentence should be tagged as English because it is used as a verb. It is transliterated to English in latter sentence and used as determiner (`this') so it was tagged as Hindi. So these kind of words were tagged based on their meanings and usage in both languages. Figure 1 describes the language annotation of tweets in the corpus.
Annotation Scheme
Two example annotations are illustrated in Figure 1. First line in every annotation consists of tweet id. Each tweet is enclosed within <tweet></tweet> tags and each word is enclosed within <word lang=“ ”></word> containing the language annotation for each word. Last line determines the category in which the tweet belongs i.e. humorous or non-humorous, enclosed by <class></class> tags. The annotated dataset with the classification system is made available online .
Error Analysis
Social media users often make spelling mistakes or use multiple variations of a word while posting a text. We replaced all such errors with their correct version. For words in Hindi that were transliterated to Latin script, we adopted a common spelling for all those words across the corpus. For example, `dis' is often used as a short form for `this', so we replaced every occurrence of `dis' to `this' in the corpus. Some examples of spelling variations are, `pese' for `paisa' (unit of curreny), `h' for `hai' (is), `ad' for `add' (addition).
System Architecture
In this section, we describe our machine learning model which is trained and tested on the corpus described in the previous section.
Pre-processing of Corpus
Preprocessing starts with tokenization, which involves separation of words using space as the delimiter and then converting the words to lower cases which is followed by the removal of punctuation marks from the tokenized tweet. All the hashtags, mentions and urls, are stored and converted to `hashtag', `mention' and `url' respectively. Sometimes hashtags provide some degree of humor in tweets, hence we segregated hashtags on the basis of camel cases and included the tokens in the tokenized tweets (hashtag decomposition) BIBREF14 , BIBREF15 . for example, #AadabArzHai can be decomposed into 3 words, `Aadab', `Arz' and `Hai'. Finally the tokenized tweets are stored along with the presence of humor as the target class.
Classification Features
The features used to build attribute vectors for training our classification model are described below. We use character level and word level features for the classification BIBREF15 . For all the features, we separated the words in the tweets based on the language annotation (Section 2.3) and prepared the feature vector for each tweet by combining the vectors for both the languages .
Previous researches shows that letter n-grams are very efficient for classifying text. They are language independent and does not require expensive text pre-processing techniques like tokenization, stemming and stop words removal, hence in the case of code-mix texts, this could yield good results BIBREF16 , BIBREF17 . Since the number of n-grams can be very large we took trigrams which occur more than ten times in the corpus.
For classifying humor in texts, it is important to understand the semantics of the sentence. Thus, we took a three word window as a feature to train our classification models to incorporate the contextual information.
Many jokes and idioms sometimes have common words. We identified those words and took them as as a feature for classification. In the preprocessing step, we decomposed hashtags using camel cases and added them along with the words. Hence, common words in the hashtags were also included in the feature vector.
Classification Approach and Results
We experimented with four different classifiers, namely, support vector machine BIBREF18 , random forest, extra tree and naive bayes classifier BIBREF19 . Chi square feature selection algorithm is applied to reduces the size of our feature vector. For training our system classifier, we used Scikit-learn BIBREF19 .
10-fold cross validation on 3543 code-mixed tweets was carried out by dividing the corpus into 10 equal parts with nine parts as training corpus and rest one for testing. Mean accuracy is calculated by taking the average of the accuracy obtained in each iteration of the testing process. Table 2 shows the accuracy for each feature when trained using mentioned classifiers along with the accuracy when all the features are used along with the overall accuracy. Support vector machine with radial basis function kernel and extra tree classifier performs better than other classifiers and yields 69.3% and 67.8% accuracy respectively. The reason kernel SVM yields the best result is because the number of observations is greator than the number of features BIBREF20 . N-grams proved to be the most efficient in all classification models followed by common words and hastags. Bag-of-words feature performed the worst in SVM, random forest and extra tree classifier but yielded better result in naive bayes classifiers. Accuracies mentioned in table 2 were calculated using fine tuning of model parameters using grid search.
Conclusion and Future Work
In this paper, we describe a freely available corpus of 3453 English-Hindi code-mixed tweets. The tweets are annotated with humorous(H) and non-humorous(N) tags along with the language tags at the word level. The task of humor identification in social media texts is analyzed as a classification problem and several machine learning classification models are used. The features used in our classification system are n-grams, bag-of-words, common words and hashtags. N-grams when trained with support vector machines with radial basis function kernel performed better than other features and yielded an accuracy of 68.5%. The best accuracy (69.3%) was given by support vector machines with radial basis function kernel.
This paper describes the initial efforts in automatic humor detection in code-mixed social media texts. Corpus can be annotated with part-of-speech tags at the word level which may yield better results in language detection. Moreover, the dataset can be further extended to include tweets from other domains. Code-mixing is very common phenomenon on social media and it is prevalent mostly in multilingual regions. It would be interesting to experiment with code-mixed texts consisting of more than two languages in which the issue of transliteration exists like Arabic, Greek and South Asian languages. Comparing training with code-mixed tweets with training with a merged corpus of monolingual tweets in English and Hindi could be an interesting future work.
Question:
What type of system does the baseline classification use?
Answer:
| support vector machine BIBREF18 , random forest, extra tree and naive bayes classifier BIBREF19 | 328 | 17,157 | 17,159 | 17,226 | ... [The rest of the article is omitted]
|
dea9e7fe8e47da5e7f31d9b1a46ebe34e731a596 | dea9e7fe8e47da5e7f31d9b1a46ebe34e731a596_1 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
“Laughter is the best Medicine” is a saying which is popular with most of the people. Humor is a form of communication that bridges the gap between various languages, cultures, ages and demographics. That's why humorous content with funny and witty hashtags are so much popular on social media. It is a very powerful tool to connect with the audience. Automatic Humor Recognition is the task of determining whether a text contains some level of humorous content or not. First conference on Computational humor was organized in 1996, since then many research have been done in this field. kao2016computational does pun detection in one-liners and dehumor detects humor in Yelp reviews. Because of the complex and interesting aspects involved in detecting humor in texts, it is one of the challenging research field in Natural Language Processing BIBREF3 . Identifying humor in a sentence sometimes require a great amount of external knowledge to completely understand it. There are many types of humor, namely anecdotes, fantasy, insult, irony, jokes, quote, self deprecation etc BIBREF4 , BIBREF5 . Most of the times there are different meanings hidden inside a sentence which is grasped differently by individuals, making the task of humor identification difficult, which is why the development of a generalized algorithm to classify different type of humor is a challenging task.
Majority of the researches on social media texts is focused on English. A study by schroeder2010half shows that, a high percentage of these texts are in non-English languages. fischer2011language gives some interesting information about the languages used on Twitter based on the geographical locations. With a huge amount of such user generated data available on social media, there is a need to develop technologies for non-English languages. In multilingual regions like South Asia, majority of the social media users speak more than two languages. In India, Hindi is the most spoken language (spoken by 41% of the population) and English is the official language of the country. Twitter has around 23.2 million monthly active users in India. Native speakers of Hindi often put English words in the sentences and transliterate the whole sentence to Latin script while posting on social media, thereby making the task of automatic text classification a very challenging problem. Linguists came up with a term for any type of language mixing, known as `code-mixing' or `code-switching' BIBREF6 , BIBREF7 , BIBREF8 , BIBREF9 . Both the terms are used interchangeably, but there is a slight difference between the two terms. Code-mixing refers to the insertion of words, phrases, and morphemes of one language into a statement or an expression of another language, whereas transliteration of every word in a sentence to another script ( here Devanagari to Latin) is coined code-switching BIBREF10 . The first tweet in Figure 1 is an example of code-mixing and second is an example of code-switching. In this paper, we use code-mixing to denote both cases.
In this paper, we present a freely available corpus containing code-mixed tweets in Hindi and English language with tweets written in Latin script. Tweets are manually classified into humorous and non-humorous classes. Moreover, each token in the tweets is also given a language tag which determines the source or origin language of the token (English or Hindi). The paper is divided in sections as follows, we start by describing the corpus and the annotation scheme in Section 2. Section 3 summarizes our supervised classification system which includes pre-processing of the tweets in the dataset and the feature extraction followed by the method used to identify humor in tweets. In the next subsection, we describe the classification model and the results of the experiments conducted using character and word level features. In the last section, we conclude the paper followed by future work and references.
Corpus Creation and Annotation
In this section, we explain the techniques used in the creation and annotation of the corpus.
Data Collection
Python package twitterscraper is used to scrap tweets from twitter. 10,478 tweets from the past two years from domains like `sports', `politics', `entertainment' were extracted. Among those tweets, we manually removed the tweets which were written either in English or Hindi entirely. There were 4161 tweets written in English and 2774 written in Hindi. Finally, a total of 3543 English-Hindi code-mixed tweets were collected. Table 1 describes the number of tweets and words in each category.
We tried to make the corpus balanced i.e. uniform distribution of tweets in each category to yield better supervised classification results as described by BIBREF11 .
Humor Annotation
The final code-mixed tweets were forwarded to a group of three annotators who were university students and fluent in both English and Hindi. Approximately 60 hours were spent in tagging tweets for the presence of humor. Tweets which consisted of any anecdotes, fantasy, irony, jokes, insults were annotated as humorous whereas tweets stating any facts, dialogues or speech which did not contain amusement were put in non-humorous class. Following are some examples of code-mixed tweets in the corpus:
“For #WontGiveItBack to work, Dhoni needs to say `Trophy toh ghar par hi bhul aaye' ”
(For #WontGiveItBack to work, Dhoni needs to say “We forgot the trophy at home”) .
Na sadak na naukri bas badh rahi gundagardi #FailedCMNitish
(No roads no naukri, only hooliganism increasing #FailedCMNitish )
Subha ka bhula agar sham ko wapas ghar aa jaye then we must thank GPS technology.
(If someone lost in the morning return home in the evening then we must thank GPS technology.)
She : Gori hai kalaiyan pehna de mujhe hari hari chudiya He *gets green bangles* She : Not this green ya, bottle green color.
(She : (sings a song) My wrists are so beautiful, give me green green bangles He *gets green bangles* She : Not this green ya, bottle green color. )
Dard dilon ke kam ho jaate... twitter par agar poetries kam ho jaate
(pain in the heart will reduce...if number of poetries decreases on twitter).
Annotators were given certain guidelines to decide whether a tweet was humorous or not. The context of the tweet could be found by searching about hashtag or keywords used in the tweet. Example (1) uses a hashtag `#WontGiveItBack' which was trending during the ICC cricket world cup 2015. Searching it on Google gave 435k results and the time of the tweet was after the final match of the tournament. So there is an observational humor in (1) as India won the world cup in 2011 and lost in 2015 , hence the tweet was classified as humorous. Any tweets stating any facts, news or reality were classified as non-humorous. There were many tweets which did not contain any hashtags, to understand the context of such tweets annotators selected some keywords from the tweet and searched them online. Example (2) contains a comment towards a political leader towards development and was categorized as non-humorous. Tweets containing normal jokes and funny quotes like in (3) and (4) were put in humorous category. There were some tweets like (5) which consists of poem or lines of a song but modified. Annotators were guided that if such tweets contains satire or any humoristic features, then it could be categorized as humorous otherwise not. There were some tweets which were typical to categorize like (5), hence it was left to the annotators to the best of their understanding. Based on the above guidelines annotators categorized the tweets. To measure inter annotator agreement we opted for Fleiss' Kappa BIBREF12 obtaining an agreement of 0.821 . Both humorous and non-humorous tweets in nearly balanced amount were selected to prepare the corpus. If we had included humorous tweets from one domain like sports and non humorous tweets from another domain like news then, it would have given high performance of classification BIBREF13 . To classify based on the semantics and not on the domain differences, we included both types of tweets from different domains. Many tweets contains a picture along with a caption. Sometimes a caption may not contain humor but combined with the picture, it can provide some degree of humor. Such tweets were removed from the corpus to make the corpus unimodal. In Figure 1, the first tweet, “Anurag Kashyap can never join AAP because ministers took oath `main kisi Anurag aur dwesh ke bina kaam karunga' ” (Anurag Kashyap can never join AAP because ministers took oath `I will work without any affection (Anurag in Hindi) and without hesitation (dwesh in Hindi)'), was classified as humorous. The second tweet, “#SakshiMalik take a bow! #proudIndian #Rio #Olympics #BronzeMedal #girlpower Hamaara khaata khul Gaya!” (#SakshiMalik take a bow! #proudIndian #Rio #Olympics #BronzeMedal #girlpower Our account opened!) was classified as non-humorous as it contains a pride statement.
Language Annotation
Code-mixing provides some challenges for language identification like spelling variations and non-grammatical structures in a sentence BIBREF0 . Therefore, we annotated the tweets with the language at the word level. Native speakers of Hindi and proficient in English, labelled the language of the tokens in the tweets. Three types of tags were assigned to the tokens , En is assigned to the tokens present in English vocabulary like “family”, “Children” etc. Similarly, Hi is assigned to the tokens present in Hindi vocabulary but transliterated to Latin script like “samay” (time), “aamaadmi” (common man). Rest of the tokens consists of proper nouns, numbers, dates, urls, hashtags, mentions, emojis and punctuations which are labelled as Ot(others). Major concern in language annotation was to annotate the words present in both languages (ambiguous words). For example, `to' (`but' in Hindi) and `is' (`this' in Hindi) , for this scenario annotators understood the context of the tweet and based on that the words were being annotated. For example, consider two sentences `This place is 500 years old' and `Is bar Modi Sarkar'(This time Modi Government). So `is' in first sentence should be tagged as English because it is used as a verb. It is transliterated to English in latter sentence and used as determiner (`this') so it was tagged as Hindi. So these kind of words were tagged based on their meanings and usage in both languages. Figure 1 describes the language annotation of tweets in the corpus.
Annotation Scheme
Two example annotations are illustrated in Figure 1. First line in every annotation consists of tweet id. Each tweet is enclosed within <tweet></tweet> tags and each word is enclosed within <word lang=“ ”></word> containing the language annotation for each word. Last line determines the category in which the tweet belongs i.e. humorous or non-humorous, enclosed by <class></class> tags. The annotated dataset with the classification system is made available online .
Error Analysis
Social media users often make spelling mistakes or use multiple variations of a word while posting a text. We replaced all such errors with their correct version. For words in Hindi that were transliterated to Latin script, we adopted a common spelling for all those words across the corpus. For example, `dis' is often used as a short form for `this', so we replaced every occurrence of `dis' to `this' in the corpus. Some examples of spelling variations are, `pese' for `paisa' (unit of curreny), `h' for `hai' (is), `ad' for `add' (addition).
System Architecture
In this section, we describe our machine learning model which is trained and tested on the corpus described in the previous section.
Pre-processing of Corpus
Preprocessing starts with tokenization, which involves separation of words using space as the delimiter and then converting the words to lower cases which is followed by the removal of punctuation marks from the tokenized tweet. All the hashtags, mentions and urls, are stored and converted to `hashtag', `mention' and `url' respectively. Sometimes hashtags provide some degree of humor in tweets, hence we segregated hashtags on the basis of camel cases and included the tokens in the tokenized tweets (hashtag decomposition) BIBREF14 , BIBREF15 . for example, #AadabArzHai can be decomposed into 3 words, `Aadab', `Arz' and `Hai'. Finally the tokenized tweets are stored along with the presence of humor as the target class.
Classification Features
The features used to build attribute vectors for training our classification model are described below. We use character level and word level features for the classification BIBREF15 . For all the features, we separated the words in the tweets based on the language annotation (Section 2.3) and prepared the feature vector for each tweet by combining the vectors for both the languages .
Previous researches shows that letter n-grams are very efficient for classifying text. They are language independent and does not require expensive text pre-processing techniques like tokenization, stemming and stop words removal, hence in the case of code-mix texts, this could yield good results BIBREF16 , BIBREF17 . Since the number of n-grams can be very large we took trigrams which occur more than ten times in the corpus.
For classifying humor in texts, it is important to understand the semantics of the sentence. Thus, we took a three word window as a feature to train our classification models to incorporate the contextual information.
Many jokes and idioms sometimes have common words. We identified those words and took them as as a feature for classification. In the preprocessing step, we decomposed hashtags using camel cases and added them along with the words. Hence, common words in the hashtags were also included in the feature vector.
Classification Approach and Results
We experimented with four different classifiers, namely, support vector machine BIBREF18 , random forest, extra tree and naive bayes classifier BIBREF19 . Chi square feature selection algorithm is applied to reduces the size of our feature vector. For training our system classifier, we used Scikit-learn BIBREF19 .
10-fold cross validation on 3543 code-mixed tweets was carried out by dividing the corpus into 10 equal parts with nine parts as training corpus and rest one for testing. Mean accuracy is calculated by taking the average of the accuracy obtained in each iteration of the testing process. Table 2 shows the accuracy for each feature when trained using mentioned classifiers along with the accuracy when all the features are used along with the overall accuracy. Support vector machine with radial basis function kernel and extra tree classifier performs better than other classifiers and yields 69.3% and 67.8% accuracy respectively. The reason kernel SVM yields the best result is because the number of observations is greator than the number of features BIBREF20 . N-grams proved to be the most efficient in all classification models followed by common words and hastags. Bag-of-words feature performed the worst in SVM, random forest and extra tree classifier but yielded better result in naive bayes classifiers. Accuracies mentioned in table 2 were calculated using fine tuning of model parameters using grid search.
Conclusion and Future Work
In this paper, we describe a freely available corpus of 3453 English-Hindi code-mixed tweets. The tweets are annotated with humorous(H) and non-humorous(N) tags along with the language tags at the word level. The task of humor identification in social media texts is analyzed as a classification problem and several machine learning classification models are used. The features used in our classification system are n-grams, bag-of-words, common words and hashtags. N-grams when trained with support vector machines with radial basis function kernel performed better than other features and yielded an accuracy of 68.5%. The best accuracy (69.3%) was given by support vector machines with radial basis function kernel.
This paper describes the initial efforts in automatic humor detection in code-mixed social media texts. Corpus can be annotated with part-of-speech tags at the word level which may yield better results in language detection. Moreover, the dataset can be further extended to include tweets from other domains. Code-mixing is very common phenomenon on social media and it is prevalent mostly in multilingual regions. It would be interesting to experiment with code-mixed texts consisting of more than two languages in which the issue of transliteration exists like Arabic, Greek and South Asian languages. Comparing training with code-mixed tweets with training with a merged corpus of monolingual tweets in English and Hindi could be an interesting future work.
Question:
What type of system does the baseline classification use?
Answer:
| Classification system use n-grams, bag-of-words, common words and hashtags as features and SVM, random forest, extra tree and NB classifiers. | 328 | 17,157 | 17,159 | 17,226 | ... [The rest of the article is omitted]
|
3355918bbdccac644afe441f085d0ffbbad565d7 | 3355918bbdccac644afe441f085d0ffbbad565d7_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Sentiment analysis has recently been one of the hottest topics in natural language processing (NLP). It is used to identify and categorise opinions expressed by reviewers on a topic or an entity. Sentiment analysis can be leveraged in marketing, social media analysis, and customer service. Although many studies have been conducted for sentiment analysis in widely spoken languages, this topic is still immature for Turkish and many other languages.
Neural networks outperform the conventional machine learning algorithms in most classification tasks, including sentiment analysis BIBREF0. In these networks, word embedding vectors are fed as input to overcome the data sparsity problem and make the representations of words more “meaningful” and robust. Those embeddings indicate how close the words are to each other in the vector space model (VSM).
Most of the studies utilise embeddings, such as word2vec BIBREF1, which take into account the syntactic and semantic representations of the words only. Discarding the sentimental aspects of words may lead to words of different polarities being close to each other in the VSM, if they share similar semantic and syntactic features.
For Turkish, there are only a few studies which leverage sentimental information in generating the word and document embeddings. Unlike the studies conducted for English and other widely-spoken languages, in this paper, we use the official dictionaries for this language and combine the unsupervised and supervised scores to generate a unified score for each dimension of the word embeddings in this task.
Our main contribution is to create original and effective word vectors that capture syntactic, semantic, and sentimental characteristics of words, and use all of this knowledge in generating embeddings. We also utilise the word2vec embeddings trained on a large corpus. Besides using these word embeddings, we also generate hand-crafted features on a review basis and create document vectors. We evaluate those embeddings on two datasets. The results show that we outperform the approaches which do not take into account the sentimental information. We also had better performances than other studies carried out on sentiment analysis in Turkish media. We also evaluated our novel embedding approaches on two English corpora of different genres. We outperformed the baseline approaches for this language as well. The source code and datasets are publicly available.
The paper is organised as follows. In Section 2, we present the existing works on sentiment classification. In Section 3, we describe the methods proposed in this work. The experimental results are shown and the main contributions of our proposed approach are discussed in Section 4. In Section 5, we conclude the paper.
Related Work
In the literature, the main consensus is that the use of dense word embeddings outperform the sparse embeddings in many tasks. Latent semantic analysis (LSA) used to be the most popular method in generating word embeddings before the invention of the word2vec and other word vector algorithms which are mostly created by shallow neural network models. Although many studies have been employed on generating word vectors including both semantic and sentimental components, generating and analysing the effects of different types of embeddings on different tasks is an emerging field for Turkish.
Latent Dirichlet allocation (LDA) is used in BIBREF2 to extract mixture of latent topics. However, it focusses on finding the latent topics of a document, not the word meanings themselves. In BIBREF3, LSA is utilised to generate word vectors, leveraging indirect co-occurrence statistics. These outperform the use of sparse vectors BIBREF4.
Some of the prior studies have also taken into account the sentimental characteristics of a word when creating word vectors BIBREF5, BIBREF6, BIBREF7. A model with semantic and sentiment components is built in BIBREF8, making use of star-ratings of reviews. In BIBREF9, a sentiment lexicon is induced preferring the use of domain-specific co-occurrence statistics over the word2vec method and they outperform the latter.
In a recent work on sentiment analysis in Turkish BIBREF10, they learn embeddings using Turkish social media. They use the word2vec algorithm, create several unsupervised hand-crafted features, generate document vectors and feed them as input into the support vector machines (SVM) approach. We outperform this baseline approach using more effective word embeddings and supervised hand-crafted features.
In English, much of the recent work on learning sentiment-specific embeddings relies only on distant supervision. In BIBREF11, emojis are used as features and a bi-LSTM (bidirectional long short-term memory) neural network model is built to learn sentiment-aware word embeddings. In BIBREF12, a neural network that learns word embeddings is built by using contextual information about the data and supervised scores of the words. This work captures the supervised information by utilising emoticons as features. Most of our approaches do not rely on a neural network model in learning embeddings. However, they produce state-of-the-art results.
Methodology
We generate several word vectors, which capture the sentimental, lexical, and contextual characteristics of words. In addition to these mostly original vectors, we also create word2vec embeddings to represent the corpus words by training the embedding model on these datasets. After generating these, we combine them with hand-crafted features to create document vectors and perform classification, as will be explained in Section 3.5.
Methodology ::: Corpus-based Approach
Contextual information is informative in the sense that, in general, similar words tend to appear in the same contexts. For example, the word smart is more likely to cooccur with the word hardworking than with the word lazy. This similarity can be defined semantically and sentimentally. In the corpus-based approach, we capture both of these characteristics and generate word embeddings specific to a domain.
Firstly, we construct a matrix whose entries correspond to the number of cooccurrences of the row and column words in sliding windows. Diagonal entries are assigned the number of sliding windows that the corresponding row word appears in the whole corpus. We then normalise each row by dividing entries in the row by the maximum score in it.
Secondly, we perform the principal component analysis (PCA) method to reduce the dimensionality. It captures latent meanings and takes into account high-order cooccurrence removing noise. The attribute (column) number of the matrix is reduced to 200. We then compute cosine similarity between each row pair $w_i$ and $w_j$ as in (DISPLAY_FORM3) to find out how similar two word vectors (rows) are.
Thirdly, all the values in the matrix are subtracted from 1 to create a dissimilarity matrix. Then, we feed the matrix as input into the fuzzy c-means clustering algorithm. We chose the number of clusters as 200, as it is considered a standard for word embeddings in the literature. After clustering, the dimension i for a corresponding word indicates the degree to which this word belongs to cluster i. The intuition behind this idea is that if two words are similar in the VSM, they are more likely to belong to the same clusters with akin probabilities. In the end, each word in the corpus is represented by a 200-dimensional vector.
In addition to this method, we also perform singular value decomposition (SVD) on the cooccurrence matrices, where we compute the matrix $M^{PPMI} = U\Sigma V^{T}$. Positive pointwise mutual information (PPMI) scores between words are calculated and the truncated singular value decomposition is computed. We take into account the U matrix only for each word. We have chosen the singular value number as 200. That is, each word in the corpus is represented by a 200-dimensional vector as follows.
Methodology ::: Dictionary-based Approach
In Turkish, there do not exist well-established sentiment lexicons as in English. In this approach, we made use of the TDK (Türk Dil Kurumu - “Turkish Language Institution”) dictionary to obtain word polarities. Although it is not a sentiment lexicon, combining it with domain-specific polarity scores obtained from the corpus led us to have state-of-the-art results.
We first construct a matrix whose row entries are corpus words and column entries are the words in their dictionary definitions. We followed the boolean approach. For instance, for the word cat, the column words occurring in its dictionary definition are given a score of 1. Those column words not appearing in the definition of cat are assigned a score of 0 for that corresponding row entry.
When we performed clustering on this matrix, we observed that those words having similar meanings are, in general, assigned to the same clusters. However, this similarity fails in capturing the sentimental characteristics. For instance, the words happy and unhappy are assigned to the same cluster, since they have the same words, such as feeling, in their dictionary definitions. However, they are of opposite polarities and should be discerned from each other.
Therefore, we utilise a metric to move such words away from each other in the VSM, even though they have common words in their dictionary definitions. We multiply each value in a row with the corresponding row word's raw supervised score, thereby having more meaningful clusters. Using the training data only, the supervised polarity score per word is calculated as in (DISPLAY_FORM4).
Here, $ w_{t}$ denotes the sentiment score of word $t$, $N_{t}$ is the number of documents (reviews or tweets) in which $t$ occurs in the dataset of positive polarity, $N$ is the number of all the words in the corpus of positive polarity. $N^{\prime }$ denotes the corpus of negative polarity. $N^{\prime }_{t}$ and $N^{\prime }$ denote similar values for the negative polarity corpus. We perform normalisation to prevent the imbalance problem and add a small number to both numerator and denominator for smoothing.
As an alternative to multiplying with the supervised polarity scores, we also separately multiplied all the row scores with only +1 if the row word is a positive word, and with -1 if it is a negative word. We have observed it boosts the performance more compared to using raw scores.
The effect of this multiplication is exemplified in Figure FIGREF7, showing the positions of word vectors in the VSM. Those “x" words are sentimentally negative words, those “o" words are sentimentally positive ones. On the top coordinate plane, the words of opposite polarities are found to be close to each other, since they have common words in their dictionary definitions. Only the information concerned with the dictionary definitions are used there, discarding the polarity scores. However, when we utilise the supervised score (+1 or -1), words of opposite polarities (e.g. “happy" and “unhappy") get far away from each other as they are translated across coordinate regions. Positive words now appear in quadrant 1, whereas negative words appear in quadrant 3. Thus, in the VSM, words that are sentimentally similar to each other could be clustered more accurately. Besides clustering, we also employed the SVD method to perform dimensionality reduction on the unsupervised dictionary algorithm and used the newly generated matrix by combining it with other subapproaches. The number of dimensions is chosen as 200 again according to the $U$ matrix. The details are given in Section 3.4. When using and evaluating this subapproach on the English corpora, we used the SentiWordNet lexicon BIBREF13. We have achieved better results for the dictionary-based algorithm when we employed the SVD reduction method compared to the use of clustering.
Methodology ::: Supervised Contextual 4-scores
Our last component is a simple metric that uses four supervised scores for each word in the corpus. We extract these scores as follows. For a target word in the corpus, we scan through all of its contexts. In addition to the target word's polarity score (the self score), out of all the polarity scores of words occurring in the same contexts as the target word, minimum, maximum, and average scores are taken into consideration. The word polarity scores are computed using (DISPLAY_FORM4). Here, we obtain those scores from the training data.
The intuition behind this method is that those four scores are more indicative of a word's polarity rather than only one (the self score). This approach is fully supervised unlike the previous two approaches.
Methodology ::: Combination of the Word Embeddings
In addition to using the three approaches independently, we also combined all the matrices generated in the previous approaches. That is, we concatenate the reduced forms (SVD - U) of corpus-based, dictionary-based, and the whole of 4-score vectors of each word, horizontally. Accordingly, each corpus word is represented by a 404-dimensional vector, since corpus-based and dictionary-based vector components are each composed of 200 dimensions, whereas the 4-score vector component is formed by four values.
The main intuition behind the ensemble method is that some approaches compensate for what the others may lack. For example, the corpus-based approach captures the domain-specific, semantic, and syntactic characteristics. On the other hand, the 4-scores method captures supervised features, and the dictionary-based approach is helpful in capturing the general semantic characteristics. That is, combining those three approaches makes word vectors more representative.
Methodology ::: Generating Document Vectors
After creating several embeddings as mentioned above, we create document (review or tweet) vectors. For each document, we sum all the vectors of words occurring in that document and take their average. In addition to it, we extract three hand-crafted polarity scores, which are minimum, mean, and maximum polarity scores, from each review. These polarity scores of words are computed as in (DISPLAY_FORM4). For example, if a review consists of five words, it would have five polarity scores and we utilise only three of these sentiment scores as mentioned. Lastly, we concatenate these three scores to the averaged word vector per review.
That is, each review is represented by the average word vector of its constituent word embeddings and three supervised scores. We then feed these inputs into the SVM approach. The flowchart of our framework is given in Figure FIGREF11. When combining the unsupervised features, which are word vectors created on a word basis, with supervised three scores extracted on a review basis, we have better state-of-the-art results.
Datasets
We utilised two datasets for both Turkish and English to evaluate our methods.
For Turkish, as the first dataset, we utilised the movie reviews which are collected from a popular website. The number of reviews in this movie corpus is 20,244 and the average number of words in reviews is 39. Each of these reviews has a star-rating score which is indicative of sentiment. These polarity scores are between the values 0.5 and 5, at intervals of 0.5. We consider a review to be negative it the score is equal to or lower than 2.5. On the other hand, if it is equal to or higher than 4, it is assumed to be positive. We have randomly selected 7,020 negative and 7,020 positive reviews and processed only them.
The second Turkish dataset is the Twitter corpus which is formed of tweets about Turkish mobile network operators. Those tweets are mostly much noisier and shorter compared to the reviews in the movie corpus. In total, there are 1,716 tweets. 973 of them are negative and 743 of them are positive. These tweets are manually annotated by two humans, where the labels are either positive or negative. We measured the Cohen's Kappa inter-annotator agreement score to be 0.82. If there was a disagreement on the polarity of a tweet, we removed it.
We also utilised two other datasets in English to test the cross-linguality of our approaches. One of them is a movie corpus collected from the web. There are 5,331 positive reviews and 5,331 negative reviews in this corpus. The other is a Twitter dataset, which has nearly 1.6 million tweets annotated through a distant supervised method BIBREF14. These tweets have positive, neutral, and negative labels. We have selected 7,020 positive tweets and 7,020 negative tweets randomly to generate a balanced dataset.
Experiments ::: Preprocessing
In Turkish, people sometimes prefer to spell English characters for the corresponding Turkish characters (e.g. i for ı, c for ç) when writing in electronic format. To normalise such words, we used the Zemberek tool BIBREF15. All punctuation marks except “!" and “?" are removed, since they do not contribute much to the polarity of a document. We took into account emoticons, such as “:))", and idioms, such as “kafayı yemek” (lose one's mind), since two or more words can express a sentiment together, irrespective of the individual words thereof. Since Turkish is an agglutinative language, we used the morphological parser and disambiguator tools BIBREF16, BIBREF17. We also performed negation handling and stop-word elimination. In negation handling, we append an underscore to the end of a word if it is negated. For example, “güzel değil" (not beautiful) is redefined as “güzel_" (beautiful_) in the feature selection stage when supervised scores are being computed.
Experiments ::: Hyperparameters
We used the LibSVM utility of the WEKA tool. We chose the linear kernel option to classify the reviews. We trained word2vec embeddings on all the four corpora using the Gensim library BIBREF18 with the skip-gram method. The dimension size of these embeddings are set at 200. As mentioned, other embeddings, which are generated utilising the clustering and the SVD approach, are also of size 200. For c-mean clustering, we set the maximum number of iterations at 25, unless it converges.
Experiments ::: Results
We evaluated our models on four corpora, which are the movie and the Twitter datasets in Turkish and English. All of the embeddings are learnt on four corpora separately. We have used the accuracy metric since all the datasets are completely or nearly completely balanced. We performed 10-fold cross-validation for both of the datasets. We used the approximate randomisation technique to test whether our results are statistically significant. Here, we tried to predict the labels of reviews and assess the performance.
We obtained varying accuracies as shown in Table TABREF17. “3 feats" features are those hand-crafted features we extracted, which are the minimum, mean, and maximum polarity scores of the reviews as explained in Section 3.5. As can be seen, at least one of our methods outperforms the baseline word2vec approach for all the Turkish and English corpora, and all categories. All of our approaches performed better when we used the supervised scores, which are extracted on a review basis, and concatenated them to word vectors. Mostly, the supervised 4-scores feature leads to the highest accuracies, since it employs the annotational information concerned with polarities on a word basis.
As can be seen in Table TABREF17, the clustering method, in general, yields the lowest scores. We found out that the corpus - SVD metric does always perform better than the clustering method. We attribute it to that in SVD the most important singular values are taken into account. The corpus - SVD technique outperforms the word2vec algorithm for some corpora. When we do not take into account the 3-feats technique, the corpus-based SVD method yields the highest accuracies for the English Twitter dataset. We show that simple models can outperform more complex models, such as the concatenation of the three subapproaches or the word2vec algorithm. Another interesting finding is that for some cases the accuracy decreases when we utilise the polarity labels, as in the case for the English Twitter dataset.
Since the TDK dictionary covers most of the domain-specific vocabulary used in the movie reviews, the dictionary method performs well. However, the dictionary lacks many of the words, occurring in the tweets; therefore, its performance is not the best of all. When the TDK method is combined with the 3-feats technique, we observed a great improvement, as can be expected.
Success rates obtained for the movie corpus are much better than those for the Twitter dataset for most of our approaches, since tweets are, in general, much shorter and noisier. We also found out that, when choosing the p value as 0.05, our results are statistically significant compared to the baseline approach in Turkish BIBREF10. Some of our subapproaches also produce better success rates than those sentiment analysis models employed in English BIBREF11, BIBREF12. We have achieved state-of-the-art results for the sentiment classification task for both Turkish and English. As mentioned, our approaches, in general, perform best in predicting the labels of reviews when three supervised scores are additionality utilised.
We also employed the convolutional neural network model (CNN). However, the SVM classifier, which is a conventional machine learning algorithm, performed better. We did not include the performances of CNN for embedding types here due to the page limit of the paper.
As a qualitative assessment of the word representations, given some query words we visualised the most similar words to those words using the cosine similarity metric. By assessing the similarities between a word and all the other corpus words, we can find the most akin words according to different approaches. Table TABREF18 shows the most similar words to given query words. Those words which are indicative of sentiment are, in general, found to be most similar to those words of the same polarity. For example, the most akin word to muhteşem (gorgeous) is 10/10, both of which have positive polarity. As can be seen in Table TABREF18, our corpus-based approach is more adept at capturing domain-specific features as compared to word2vec, which generally captures general semantic and syntactic characteristics, but not the sentimental ones.
Conclusion
We have demonstrated that using word vectors that capture only semantic and syntactic characteristics may be improved by taking into account their sentimental aspects as well. Our approaches are cross-lingual and cross-domain. They can be applied to other domains and other languages than Turkish and English with minor changes.
Our study is one of the few ones that perform sentiment analysis in Turkish and leverages sentimental characteristics of words in generating word vectors and outperforms all the others. Any of the approaches we propose can be used independently of the others. Our approaches without using sentiment labels can be applied to other classification tasks, such as topic classification and concept mining.
The experiments show that even unsupervised approaches, as in the corpus-based approach, can outperform supervised approaches in classification tasks. Combining some approaches, which can compensate for what others lack, can help us build better vectors. Our word vectors are created by conventional machine learning algorithms; however, they, as in the corpus-based model, produce state-of-the-art results. Although we preferred to use a classical machine learning algorithm, which is SVM, over a neural network classifier to predict the labels of reviews, we achieved accuracies of over 90 per cent for the Turkish movie corpus and about 88 per cent for the English Twitter dataset.
We performed only binary sentiment classification in this study as most of the studies in the literature do. We will extend our system in future by using neutral reviews as well. We also plan to employ Turkish WordNet to enhance the generalisability of our embeddings as another future work.
Acknowledgments
This work was supported by Boğaziçi University Research Fund Grant Number 6980D, and by Turkish Ministry of Development under the TAM Project number DPT2007K12-0610. Cem Rıfkı Aydın is supported by TÜBİTAK BIDEB 2211E.
Question:
How are the supervised scores of the words calculated?
Answer:
| (+1 or -1), words of opposite polarities (e.g. “happy" and “unhappy") get far away from each other | 328 | 24,650 | 24,652 | 24,716 | ... [The rest of the article is omitted]
|
d9980676a83295dda37c20cfd5d58e574d0a4859 | d9980676a83295dda37c20cfd5d58e574d0a4859_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
The new generation of Neural Machine Translation (NMT) systems is known to be extremely data hungry BIBREF0 . Yet, most existing NMT training pipelines fail to fully take advantage of the very large volume of monolingual source and/or parallel data that is often available. Making a better use of data is particularly critical in domain adaptation scenarios, where parallel adaptation data is usually assumed to be small in comparison to out-of-domain parallel data, or to in-domain monolingual texts. This situation sharply contrasts with the previous generation of statistical MT engines BIBREF1 , which could seamlessly integrate very large amounts of non-parallel documents, usually with a large positive effect on translation quality.
Such observations have been made repeatedly and have led to many innovative techniques to integrate monolingual data in NMT, that we review shortly. The most successful approach to date is the proposal of BIBREF2 , who use monolingual target texts to generate artificial parallel data via backward translation (BT). This technique has since proven effective in many subsequent studies. It is however very computationally costly, typically requiring to translate large sets of data. Determining the “right” amount (and quality) of BT data is another open issue, but we observe that experiments reported in the literature only use a subset of the available monolingual resources. This suggests that standard recipes for BT might be sub-optimal.
This paper aims to better understand the strengths and weaknesses of BT and to design more principled techniques to improve its effects. More specifically, we seek to answer the following questions: since there are many ways to generate pseudo parallel corpora, how important is the quality of this data for MT performance? Which properties of back-translated sentences actually matter for MT quality? Does BT act as some kind of regularizer BIBREF3 ? Can BT be efficiently simulated? Does BT data play the same role as a target-side language modeling, or are they complementary? BT is often used for domain adaptation: can the effect of having more in-domain data be sorted out from the mere increase of training material BIBREF2 ? For studies related to the impact of varying the size of BT data, we refer the readers to the recent work of BIBREF4 .
To answer these questions, we have reimplemented several strategies to use monolingual data in NMT and have run experiments on two language pairs in a very controlled setting (see § SECREF2 ). Our main results (see § SECREF4 and § SECREF5 ) suggest promising directions for efficient domain adaptation with cheaper techniques than conventional BT.
In-domain and out-of-domain data
We are mostly interested with the following training scenario: a large out-of-domain parallel corpus, and limited monolingual in-domain data. We focus here on the Europarl domain, for which we have ample data in several languages, and use as in-domain training data the Europarl corpus BIBREF5 for two translation directions: English INLINEFORM0 German and English INLINEFORM1 French. As we study the benefits of monolingual data, most of our experiments only use the target side of this corpus. The rationale for choosing this domain is to (i) to perform large scale comparisons of synthetic and natural parallel corpora; (ii) to study the effect of BT in a well-defined domain-adaptation scenario. For both language pairs, we use the Europarl tests from 2007 and 2008 for evaluation purposes, keeping test 2006 for development. When measuring out-of-domain performance, we will use the WMT newstest 2014.
NMT setups and performance
Our baseline NMT system implements the attentional encoder-decoder approach BIBREF6 , BIBREF7 as implemented in Nematus BIBREF8 on 4 million out-of-domain parallel sentences. For French we use samples from News-Commentary-11 and Wikipedia from WMT 2014 shared translation task, as well as the Multi-UN BIBREF9 and EU-Bookshop BIBREF10 corpora. For German, we use samples from News-Commentary-11, Rapid, Common-Crawl (WMT 2017) and Multi-UN (see table TABREF5 ). Bilingual BPE units BIBREF11 are learned with 50k merge operations, yielding vocabularies of about respectively 32k and 36k for English INLINEFORM0 French and 32k and 44k for English INLINEFORM1 German.
Both systems use 512-dimensional word embeddings and a single hidden layer with 1024 cells. They are optimized using Adam BIBREF12 and early stopped according to the validation performance. Training lasted for about three weeks on an Nvidia K80 GPU card.
Systems generating back-translated data are trained using the same out-of-domain corpus, where we simply exchange the source and target sides. They are further documented in § SECREF8 . For the sake of comparison, we also train a system that has access to a large batch of in-domain parallel data following the strategy often referred to as “fine-tuning”: upon convergence of the baseline model, we resume training with a 2M sentence in-domain corpus mixed with an equal amount of randomly selected out-of-domain natural sentences, with the same architecture and training parameters, running validation every 2000 updates with a patience of 10. Since BPE units are selected based only on the out-of-domain statistics, fine-tuning is performed on sentences that are slightly longer (ie. they contain more units) than for the initial training. This system defines an upper-bound of the translation performance and is denoted below as natural.
Our baseline and topline results are in Table TABREF6 , where we measure translation performance using BLEU BIBREF13 , BEER BIBREF14 (higher is better) and characTER BIBREF15 (smaller is better). As they are trained from much smaller amounts of data than current systems, these baselines are not quite competitive to today's best system, but still represent serious baselines for these datasets. Given our setups, fine-tuning with in-domain natural data improves BLEU by almost 4 points for both translation directions on in-domain tests; it also improves, albeit by a smaller margin, the BLEU score of the out-of-domain tests.
Using artificial parallel data in NMT
A simple way to use monolingual data in MT is to turn it into synthetic parallel data and let the training procedure run as usual BIBREF16 . In this section, we explore various ways to implement this strategy. We first reproduce results of BIBREF2 with BT of various qualities, that we then analyze thoroughly.
The quality of Back-Translation
BT requires the availability of an MT system in the reverse translation direction. We consider here three MT systems of increasing quality:
backtrans-bad: this is a very poor SMT system trained using only 50k parallel sentences from the out-of-domain data, and no additional monolingual data. For this system as for the next one, we use Moses BIBREF17 out-of-the-box, computing alignments with Fastalign BIBREF18 , with a minimal pre-processing (basic tokenization). This setting provides us with a pessimistic estimate of what we could get in low-resource conditions.
backtrans-good: these are much larger SMT systems, which use the same parallel data as the baseline NMTs (see § SECREF4 ) and all the English monolingual data available for the WMT 2017 shared tasks, totalling approximately 174M sentences. These systems are strong, yet relatively cheap to build.
backtrans-nmt: these are the best NMT systems we could train, using settings that replicate the forward translation NMTs.
Note that we do not use any in-domain (Europarl) data to train these systems. Their performance is reported in Table TABREF7 , where we observe a 12 BLEU points gap between the worst and best systems (for both languages).
As noted eg. in BIBREF19 , BIBREF20 , artificial parallel data obtained through forward-translation (FT) can also prove advantageous and we also consider a FT system (fwdtrans-nmt): in this case the target side of the corpus is artificial and is generated using the baseline NMT applied to a natural source.
Our results (see Table TABREF6 ) replicate the findings of BIBREF2 : large gains can be obtained from BT (nearly INLINEFORM0 BLEU in French and German); better artificial data yields better translation systems. Interestingly, our best Moses system is almost as good as the NMT and an order of magnitude faster to train. Improvements obtained with the bad system are much smaller; contrary to the better MTs, this system is even detrimental for the out-of-domain test.
Gains with forward translation are significant, as in BIBREF21 , albeit about half as good as with BT, and result in small improvements for the in-domain and for the out-of-domain tests. Experiments combining forward and backward translation (backfwdtrans-nmt), each using a half of the available artificial data, do not outperform the best BT results.
We finally note the large remaining difference between BT data and natural data, even though they only differ in their source side. This shows that at least in our domain-adaptation settings, BT does not really act as a regularizer, contrarily to the findings of BIBREF4 , BIBREF11 . Figure FIGREF13 displays the learning curves of these two systems. We observe that backtrans-nmt improves quickly in the earliest updates and then stays horizontal, whereas natural continues improving, even after 400k updates. Therefore BT does not help to avoid overfitting, it actually encourages it, which may be due “easier” training examples (cf. § SECREF15 ).
Properties of back-translated data
Comparing the natural and artificial sources of our parallel data wrt. several linguistic and distributional properties, we observe that (see Fig. FIGREF21 - FIGREF22 ):
artificial sources are on average shorter than natural ones: when using BT, cases where the source is shorter than the target are rarer; cases when they have the same length are more frequent.
automatic word alignments between artificial sources tend to be more monotonic than when using natural sources, as measured by the average Kendall INLINEFORM0 of source-target alignments BIBREF22 : for French-English the respective numbers are 0.048 (natural) and 0.018 (artificial); for German-English 0.068 and 0.053. Using more monotonic sentence pairs turns out to be a facilitating factor for NMT, as also noted by BIBREF20 .
syntactically, artificial sources are simpler than real data; We observe significant differences in the distributions of tree depths.
distributionally, plain word occurrences in artificial sources are more concentrated; this also translates into both a slower increase of the number of types wrt. the number of sentences and a smaller number of rare events.
The intuition is that properties (i) and (ii) should help translation as compared to natural source, while property (iv) should be detrimental. We checked (ii) by building systems with only 10M words from the natural parallel data selecting these data either randomly or based on the regularity of their word alignments. Results in Table TABREF23 show that the latter is much preferable for the overall performance. This might explain that the mostly monotonic BT from Moses are almost as good as the fluid BT from NMT and that both boost the baseline.
Stupid Back-Translation
We now analyze the effect of using much simpler data generation schemes, which do not require the availability of a backward translation engine.
Setups
We use the following cheap ways to generate pseudo-source texts:
copy: in this setting, the source side is a mere copy of the target-side data. Since the source vocabulary of the NMT is fixed, copying the target sentences can cause the occurrence of OOVs. To avoid this situation, BIBREF24 decompose the target words into source-side units to make the copy look like source sentences. Each OOV found in the copy is split into smaller units until all the resulting chunks are in the source vocabulary.
copy-marked: another way to integrate copies without having to deal with OOVs is to augment the source vocabulary with a copy of the target vocabulary. In this setup, BIBREF25 ensure that both vocabularies never overlap by marking the target word copies with a special language identifier. Therefore the English word resume cannot be confused with the homographic French word, which is marked @fr@resume.
copy-dummies: instead of using actual copies, we replace each word with “dummy” tokens. We use this unrealistic setup to observe the training over noisy and hardly informative source sentences.
We then use the procedures described in § SECREF4 , except that the pseudo-source embeddings in the copy-marked setup are pretrained for three epochs on the in-domain data, while all remaining parameters are frozen. This prevents random parameters from hurting the already trained model.
Copy+marking+noise is not so stupid
We observe that the copy setup has only a small impact on the English-French system, for which the baseline is already strong. This is less true for English-German where simple copies yield a significant improvement. Performance drops for both language pairs in the copy-dummies setup.
We achieve our best gains with the copy-marked setup, which is the best way to use a copy of the target (although the performance on the out-of-domain tests is at most the same as the baseline). Such gains may look surprising, since the NMT model does not need to learn to translate but only to copy the source. This is indeed what happens: to confirm this, we built a fake test set having identical source and target side (in French). The average cross-entropy for this test set is 0.33, very close to 0, to be compared with an average cost of 58.52 when we process an actual source (in English). This means that the model has learned to copy words from source to target with no difficulty, even for sentences not seen in training. A follow-up question is whether training a copying task instead of a translation task limits the improvement: would the NMT learn better if the task was harder? To measure this, we introduce noise in the target sentences copied onto the source, following the procedure of BIBREF26 : it deletes random words and performs a small random permutation of the remaining words. Results (+ Source noise) show no difference for the French in-domain test sets, but bring the out-of-domain score to the level of the baseline. Finally, we observe a significant improvement on German in-domain test sets, compared to the baseline (about +1.5 BLEU). This last setup is even almost as good as the backtrans-nmt condition (see § SECREF8 ) for German. This shows that learning to reorder and predict missing words can more effectively serve our purposes than simply learning to copy.
Towards more natural pseudo-sources
Integrating monolingual data into NMT can be as easy as copying the target into the source, which already gives some improvement; adding noise makes things even better. We now study ways to make pseudo-sources look more like natural data, using the framework of Generative Adversarial Networks (GANs) BIBREF27 , an idea borrowed from BIBREF26 .
GAN setups
In our setups, we use a marked target copy, viewed as a fake source, which a generator encodes so as to fool a discriminator trained to distinguish a fake from a natural source. Our architecture contains two distinct encoders, one for the natural source and one for the pseudo-source. The latter acts as the generator ( INLINEFORM0 ) in the GAN framework, computing a representation of the pseudo-source that is then input to a discriminator ( INLINEFORM1 ), which has to sort natural from artificial encodings. INLINEFORM2 assigns a probability of a sentence being natural.
During training, the cost of the discriminator is computed over two batches, one with natural (out-of-domain) sentences INLINEFORM0 and one with (in-domain) pseudo-sentences INLINEFORM1 . The discriminator is a bidirectional-Recurrent Neural Network (RNN) of dimension 1024. Averaged states are passed to a single feed-forward layer, to which a sigmoid is applied. It inputs encodings of natural ( INLINEFORM2 ) and pseudo-sentences ( INLINEFORM3 ) and is trained to optimize:
INLINEFORM0
INLINEFORM0 's parameters are updated to maximally fool INLINEFORM1 , thus the loss INLINEFORM2 : INLINEFORM3
Finally, we keep the usual MT objective. ( INLINEFORM0 is a real or pseudo-sentence): INLINEFORM1
We thus need to train three sets of parameters: INLINEFORM0 and INLINEFORM1 (MT parameters), with INLINEFORM2 . The pseudo-source encoder and embeddings are updated wrt. both INLINEFORM3 and INLINEFORM4 . Following BIBREF28 , INLINEFORM5 is updated only when INLINEFORM6 's accuracy exceeds INLINEFORM7 . On the other hand, INLINEFORM8 is not updated when its accuracy exceeds INLINEFORM9 . At each update, two batches are generated for each type of data, which are encoded with the real or pseudo-encoder. The encoder outputs serve to compute INLINEFORM10 and INLINEFORM11 . Finally, the pseudo-source is encoded again (once INLINEFORM12 is updated), both encoders are plugged into the translation model and the MT cost is back-propagated down to the real and pseudo-word embeddings. Pseudo-encoder and discriminator parameters are pre-trained for 10k updates. At test time, the pseudo-encoder is ignored and inference is run as usual.
GANs can help
Results are in Table TABREF32 , assuming the same fine-tuning procedure as above. On top of the copy-marked setup, our GANs do not provide any improvement in both language pairs, with the exception of a small improvement for English-French on the out-of-domain test, which we understand as a sign that the model is more robust to domain variations, just like when adding pseudo-source noise. When combined with noise, the French model yields the best performance we could obtain with stupid BT on the in-domain tests, at least in terms of BLEU and BEER. On the News domain, we remain close to the baseline level, with slight improvements in German.
A first observation is that this method brings stupid BT models closer to conventional BT, at a greatly reduced computational cost. While French still remains 0.4 to 1.0 BLEU below very good backtranslation, both approaches are in the same ballpark for German - may be because BTs are better for the former system than for the latter.
Finally note that the GAN architecture has two differences with basic copy-marked: (a) a distinct encoder for real and pseudo-sentence; (b) a different training regime for these encoders. To sort out the effects of (a) and (b), we reproduce the GAN setup with BT sentences, instead of copies. Using a separate encoder for the pseudo-source in the backtrans-nmt setup can be detrimental to performance (see Table TABREF32 ): translation degrades in French for all metrics. Adding GANs on top of the pseudo-encoder was not able to make up for the degradation observed in French, but allowed the German system to slightly outperform backtrans-nmt. Even though this setup is unrealistic and overly costly, it shows that GANs are actually helping even good systems.
Using Target Language Models
In this section, we compare the previous methods with the use of a target side Language Model (LM). Several proposals exist in the literature to integrate LMs in NMT: for instance, BIBREF3 strengthen the decoder by integrating an extra, source independent, RNN layer in a conventional NMT architecture. Training is performed either with parallel, or monolingual data. In the latter case, word prediction only relies on the source independent part of the network.
LM Setup
We have followed BIBREF29 and reimplemented their deep-fusion technique. It requires to first independently learn a RNN-LM on the in-domain target data with a cross-entropy objective; then to train the optimal combination of the translation and the language models by adding the hidden state of the RNN-LM as an additional input to the softmax layer of the decoder. Our RNN-LMs are trained using dl4mt with the target side of the parallel data and the Europarl corpus (about 6M sentences for both French and German), using a one-layer GRU with the same dimension as the MT decoder (1024).
LM Results
Results are in Table TABREF33 . They show that deep-fusion hardly improves the Europarl results, while we obtain about +0.6 BLEU over the baseline on newstest-2014 for both languages. deep-fusion differs from stupid BT in that the model is not directly optimized on the in-domain data, but uses the LM trained on Europarl to maximize the likelihood of the out-of-domain training data. Therefore, no specific improvement is to be expected in terms of domain adaptation, and the performance increases in the more general domain. Combining deep-fusion and copy-marked + noise + GANs brings slight improvements on the German in-domain test sets, and performance out of the domain remains near the baseline level.
Re-analyzing the effects of BT
As a follow up of previous discussions, we analyze the effect of BT on the internals of the network. Arguably, using a copy of the target sentence instead of a natural source should not be helpful for the encoder, but is it also the case with a strong BT? What are the effects on the attention model?
Parameter freezing protocol
To investigate these questions, we run the same fine-tuning using the copy-marked, backtrans-nmt and backtrans-nmt setups. Note that except for the last one, all training scenarios have access to same target training data. We intend to see whether the overall performance of the NMT system degrades when we selectively freeze certain sets of parameters, meaning that they are not updated during fine-tuning.
Results
BLEU scores are in Table TABREF39 . The backtrans-nmt setup is hardly impacted by selective updates: updating the only decoder leads to a degradation of at most 0.2 BLEU. For copy-marked, we were not able to freeze the source embeddings, since these are initialized when fine-tuning begins and therefore need to be trained. We observe that freezing the encoder and/or the attention parameters has no impact on the English-German system, whereas it slightly degrades the English-French one. This suggests that using artificial sources, even of the poorest quality, has a positive impact on all the components of the network, which makes another big difference with the LM integration scenario.
The largest degradation is for natural, where the model is prevented from learning from informative source sentences, which leads to a decrease of 0.4 to over 1.0 BLEU. We assume from these experiments that BT impacts most of all the decoder, and learning to encode a pseudo-source, be it a copy or an actual back-translation, only marginally helps to significantly improve the quality. Finally, in the fwdtrans-nmt setup, freezing the decoder does not seem to harm learning with a natural source.
Related work
The literature devoted to the use of monolingual data is large, and quickly expanding. We already alluded to several possible ways to use such data: using back- or forward-translation or using a target language model. The former approach is mostly documented in BIBREF2 , and recently analyzed in BIBREF19 , which focus on fully artificial settings as well as pivot-based artificial data; and BIBREF4 , which studies the effects of increasing the size of BT data. The studies of BIBREF20 , BIBREF19 also consider forward translation and BIBREF21 expand these results to domain adaptation scenarios. Our results are complementary to these earlier studies.
As shown above, many alternatives to BT exist. The most obvious is to use target LMs BIBREF3 , BIBREF29 , as we have also done here; but attempts to improve the encoder using multi-task learning also exist BIBREF30 .
This investigation is also related to recent attempts to consider supplementary data with a valid target side, such as multi-lingual NMT BIBREF31 , where source texts in several languages are fed in the same encoder-decoder architecture, with partial sharing of the layers. This is another realistic scenario where additional resources can be used to selectively improve parts of the model.
Round trip training is another important source of inspiration, as it can be viewed as a way to use BT to perform semi-unsupervised BIBREF32 or unsupervised BIBREF33 training of NMT. The most convincing attempt to date along these lines has been proposed by BIBREF26 , who propose to use GANs to mitigate the difference between artificial and natural data.
Conclusion
In this paper, we have analyzed various ways to integrate monolingual data in an NMT framework, focusing on their impact on quality and domain adaptation. While confirming the effectiveness of BT, our study also proposed significantly cheaper ways to improve the baseline performance, using a slightly modified copy of the target, instead of its full BT. When no high quality BT is available, using GANs to make the pseudo-source sentences closer to natural source sentences is an efficient solution for domain adaptation.
To recap our answers to our initial questions: the quality of BT actually matters for NMT (cf. § SECREF8 ) and it seems that, even though artificial source are lexically less diverse and syntactically complex than real sentence, their monotonicity is a facilitating factor. We have studied cheaper alternatives and found out that copies of the target, if properly noised (§ SECREF4 ), and even better, if used with GANs, could be almost as good as low quality BTs (§ SECREF5 ): BT is only worth its cost when good BT can be generated. Finally, BT seems preferable to integrating external LM - at least in our data condition (§ SECREF6 ). Further experiments with larger LMs are needed to confirm this observation, and also to evaluate the complementarity of both strategies. More work is needed to better understand the impact of BT on subparts of the network (§ SECREF7 ).
In future work, we plan to investigate other cheap ways to generate artificial data. The experimental setup we proposed may also benefit from a refining of the data selection strategies to focus on the most useful monolingual sentences.
Question:
what data simulation techniques were introduced?
Answer:
| copy, copy-marked, copy-dummies | 328 | 26,501 | 26,503 | 26,561 | ... [The rest of the article is omitted]
|
d9980676a83295dda37c20cfd5d58e574d0a4859 | d9980676a83295dda37c20cfd5d58e574d0a4859_1 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
The new generation of Neural Machine Translation (NMT) systems is known to be extremely data hungry BIBREF0 . Yet, most existing NMT training pipelines fail to fully take advantage of the very large volume of monolingual source and/or parallel data that is often available. Making a better use of data is particularly critical in domain adaptation scenarios, where parallel adaptation data is usually assumed to be small in comparison to out-of-domain parallel data, or to in-domain monolingual texts. This situation sharply contrasts with the previous generation of statistical MT engines BIBREF1 , which could seamlessly integrate very large amounts of non-parallel documents, usually with a large positive effect on translation quality.
Such observations have been made repeatedly and have led to many innovative techniques to integrate monolingual data in NMT, that we review shortly. The most successful approach to date is the proposal of BIBREF2 , who use monolingual target texts to generate artificial parallel data via backward translation (BT). This technique has since proven effective in many subsequent studies. It is however very computationally costly, typically requiring to translate large sets of data. Determining the “right” amount (and quality) of BT data is another open issue, but we observe that experiments reported in the literature only use a subset of the available monolingual resources. This suggests that standard recipes for BT might be sub-optimal.
This paper aims to better understand the strengths and weaknesses of BT and to design more principled techniques to improve its effects. More specifically, we seek to answer the following questions: since there are many ways to generate pseudo parallel corpora, how important is the quality of this data for MT performance? Which properties of back-translated sentences actually matter for MT quality? Does BT act as some kind of regularizer BIBREF3 ? Can BT be efficiently simulated? Does BT data play the same role as a target-side language modeling, or are they complementary? BT is often used for domain adaptation: can the effect of having more in-domain data be sorted out from the mere increase of training material BIBREF2 ? For studies related to the impact of varying the size of BT data, we refer the readers to the recent work of BIBREF4 .
To answer these questions, we have reimplemented several strategies to use monolingual data in NMT and have run experiments on two language pairs in a very controlled setting (see § SECREF2 ). Our main results (see § SECREF4 and § SECREF5 ) suggest promising directions for efficient domain adaptation with cheaper techniques than conventional BT.
In-domain and out-of-domain data
We are mostly interested with the following training scenario: a large out-of-domain parallel corpus, and limited monolingual in-domain data. We focus here on the Europarl domain, for which we have ample data in several languages, and use as in-domain training data the Europarl corpus BIBREF5 for two translation directions: English INLINEFORM0 German and English INLINEFORM1 French. As we study the benefits of monolingual data, most of our experiments only use the target side of this corpus. The rationale for choosing this domain is to (i) to perform large scale comparisons of synthetic and natural parallel corpora; (ii) to study the effect of BT in a well-defined domain-adaptation scenario. For both language pairs, we use the Europarl tests from 2007 and 2008 for evaluation purposes, keeping test 2006 for development. When measuring out-of-domain performance, we will use the WMT newstest 2014.
NMT setups and performance
Our baseline NMT system implements the attentional encoder-decoder approach BIBREF6 , BIBREF7 as implemented in Nematus BIBREF8 on 4 million out-of-domain parallel sentences. For French we use samples from News-Commentary-11 and Wikipedia from WMT 2014 shared translation task, as well as the Multi-UN BIBREF9 and EU-Bookshop BIBREF10 corpora. For German, we use samples from News-Commentary-11, Rapid, Common-Crawl (WMT 2017) and Multi-UN (see table TABREF5 ). Bilingual BPE units BIBREF11 are learned with 50k merge operations, yielding vocabularies of about respectively 32k and 36k for English INLINEFORM0 French and 32k and 44k for English INLINEFORM1 German.
Both systems use 512-dimensional word embeddings and a single hidden layer with 1024 cells. They are optimized using Adam BIBREF12 and early stopped according to the validation performance. Training lasted for about three weeks on an Nvidia K80 GPU card.
Systems generating back-translated data are trained using the same out-of-domain corpus, where we simply exchange the source and target sides. They are further documented in § SECREF8 . For the sake of comparison, we also train a system that has access to a large batch of in-domain parallel data following the strategy often referred to as “fine-tuning”: upon convergence of the baseline model, we resume training with a 2M sentence in-domain corpus mixed with an equal amount of randomly selected out-of-domain natural sentences, with the same architecture and training parameters, running validation every 2000 updates with a patience of 10. Since BPE units are selected based only on the out-of-domain statistics, fine-tuning is performed on sentences that are slightly longer (ie. they contain more units) than for the initial training. This system defines an upper-bound of the translation performance and is denoted below as natural.
Our baseline and topline results are in Table TABREF6 , where we measure translation performance using BLEU BIBREF13 , BEER BIBREF14 (higher is better) and characTER BIBREF15 (smaller is better). As they are trained from much smaller amounts of data than current systems, these baselines are not quite competitive to today's best system, but still represent serious baselines for these datasets. Given our setups, fine-tuning with in-domain natural data improves BLEU by almost 4 points for both translation directions on in-domain tests; it also improves, albeit by a smaller margin, the BLEU score of the out-of-domain tests.
Using artificial parallel data in NMT
A simple way to use monolingual data in MT is to turn it into synthetic parallel data and let the training procedure run as usual BIBREF16 . In this section, we explore various ways to implement this strategy. We first reproduce results of BIBREF2 with BT of various qualities, that we then analyze thoroughly.
The quality of Back-Translation
BT requires the availability of an MT system in the reverse translation direction. We consider here three MT systems of increasing quality:
backtrans-bad: this is a very poor SMT system trained using only 50k parallel sentences from the out-of-domain data, and no additional monolingual data. For this system as for the next one, we use Moses BIBREF17 out-of-the-box, computing alignments with Fastalign BIBREF18 , with a minimal pre-processing (basic tokenization). This setting provides us with a pessimistic estimate of what we could get in low-resource conditions.
backtrans-good: these are much larger SMT systems, which use the same parallel data as the baseline NMTs (see § SECREF4 ) and all the English monolingual data available for the WMT 2017 shared tasks, totalling approximately 174M sentences. These systems are strong, yet relatively cheap to build.
backtrans-nmt: these are the best NMT systems we could train, using settings that replicate the forward translation NMTs.
Note that we do not use any in-domain (Europarl) data to train these systems. Their performance is reported in Table TABREF7 , where we observe a 12 BLEU points gap between the worst and best systems (for both languages).
As noted eg. in BIBREF19 , BIBREF20 , artificial parallel data obtained through forward-translation (FT) can also prove advantageous and we also consider a FT system (fwdtrans-nmt): in this case the target side of the corpus is artificial and is generated using the baseline NMT applied to a natural source.
Our results (see Table TABREF6 ) replicate the findings of BIBREF2 : large gains can be obtained from BT (nearly INLINEFORM0 BLEU in French and German); better artificial data yields better translation systems. Interestingly, our best Moses system is almost as good as the NMT and an order of magnitude faster to train. Improvements obtained with the bad system are much smaller; contrary to the better MTs, this system is even detrimental for the out-of-domain test.
Gains with forward translation are significant, as in BIBREF21 , albeit about half as good as with BT, and result in small improvements for the in-domain and for the out-of-domain tests. Experiments combining forward and backward translation (backfwdtrans-nmt), each using a half of the available artificial data, do not outperform the best BT results.
We finally note the large remaining difference between BT data and natural data, even though they only differ in their source side. This shows that at least in our domain-adaptation settings, BT does not really act as a regularizer, contrarily to the findings of BIBREF4 , BIBREF11 . Figure FIGREF13 displays the learning curves of these two systems. We observe that backtrans-nmt improves quickly in the earliest updates and then stays horizontal, whereas natural continues improving, even after 400k updates. Therefore BT does not help to avoid overfitting, it actually encourages it, which may be due “easier” training examples (cf. § SECREF15 ).
Properties of back-translated data
Comparing the natural and artificial sources of our parallel data wrt. several linguistic and distributional properties, we observe that (see Fig. FIGREF21 - FIGREF22 ):
artificial sources are on average shorter than natural ones: when using BT, cases where the source is shorter than the target are rarer; cases when they have the same length are more frequent.
automatic word alignments between artificial sources tend to be more monotonic than when using natural sources, as measured by the average Kendall INLINEFORM0 of source-target alignments BIBREF22 : for French-English the respective numbers are 0.048 (natural) and 0.018 (artificial); for German-English 0.068 and 0.053. Using more monotonic sentence pairs turns out to be a facilitating factor for NMT, as also noted by BIBREF20 .
syntactically, artificial sources are simpler than real data; We observe significant differences in the distributions of tree depths.
distributionally, plain word occurrences in artificial sources are more concentrated; this also translates into both a slower increase of the number of types wrt. the number of sentences and a smaller number of rare events.
The intuition is that properties (i) and (ii) should help translation as compared to natural source, while property (iv) should be detrimental. We checked (ii) by building systems with only 10M words from the natural parallel data selecting these data either randomly or based on the regularity of their word alignments. Results in Table TABREF23 show that the latter is much preferable for the overall performance. This might explain that the mostly monotonic BT from Moses are almost as good as the fluid BT from NMT and that both boost the baseline.
Stupid Back-Translation
We now analyze the effect of using much simpler data generation schemes, which do not require the availability of a backward translation engine.
Setups
We use the following cheap ways to generate pseudo-source texts:
copy: in this setting, the source side is a mere copy of the target-side data. Since the source vocabulary of the NMT is fixed, copying the target sentences can cause the occurrence of OOVs. To avoid this situation, BIBREF24 decompose the target words into source-side units to make the copy look like source sentences. Each OOV found in the copy is split into smaller units until all the resulting chunks are in the source vocabulary.
copy-marked: another way to integrate copies without having to deal with OOVs is to augment the source vocabulary with a copy of the target vocabulary. In this setup, BIBREF25 ensure that both vocabularies never overlap by marking the target word copies with a special language identifier. Therefore the English word resume cannot be confused with the homographic French word, which is marked @fr@resume.
copy-dummies: instead of using actual copies, we replace each word with “dummy” tokens. We use this unrealistic setup to observe the training over noisy and hardly informative source sentences.
We then use the procedures described in § SECREF4 , except that the pseudo-source embeddings in the copy-marked setup are pretrained for three epochs on the in-domain data, while all remaining parameters are frozen. This prevents random parameters from hurting the already trained model.
Copy+marking+noise is not so stupid
We observe that the copy setup has only a small impact on the English-French system, for which the baseline is already strong. This is less true for English-German where simple copies yield a significant improvement. Performance drops for both language pairs in the copy-dummies setup.
We achieve our best gains with the copy-marked setup, which is the best way to use a copy of the target (although the performance on the out-of-domain tests is at most the same as the baseline). Such gains may look surprising, since the NMT model does not need to learn to translate but only to copy the source. This is indeed what happens: to confirm this, we built a fake test set having identical source and target side (in French). The average cross-entropy for this test set is 0.33, very close to 0, to be compared with an average cost of 58.52 when we process an actual source (in English). This means that the model has learned to copy words from source to target with no difficulty, even for sentences not seen in training. A follow-up question is whether training a copying task instead of a translation task limits the improvement: would the NMT learn better if the task was harder? To measure this, we introduce noise in the target sentences copied onto the source, following the procedure of BIBREF26 : it deletes random words and performs a small random permutation of the remaining words. Results (+ Source noise) show no difference for the French in-domain test sets, but bring the out-of-domain score to the level of the baseline. Finally, we observe a significant improvement on German in-domain test sets, compared to the baseline (about +1.5 BLEU). This last setup is even almost as good as the backtrans-nmt condition (see § SECREF8 ) for German. This shows that learning to reorder and predict missing words can more effectively serve our purposes than simply learning to copy.
Towards more natural pseudo-sources
Integrating monolingual data into NMT can be as easy as copying the target into the source, which already gives some improvement; adding noise makes things even better. We now study ways to make pseudo-sources look more like natural data, using the framework of Generative Adversarial Networks (GANs) BIBREF27 , an idea borrowed from BIBREF26 .
GAN setups
In our setups, we use a marked target copy, viewed as a fake source, which a generator encodes so as to fool a discriminator trained to distinguish a fake from a natural source. Our architecture contains two distinct encoders, one for the natural source and one for the pseudo-source. The latter acts as the generator ( INLINEFORM0 ) in the GAN framework, computing a representation of the pseudo-source that is then input to a discriminator ( INLINEFORM1 ), which has to sort natural from artificial encodings. INLINEFORM2 assigns a probability of a sentence being natural.
During training, the cost of the discriminator is computed over two batches, one with natural (out-of-domain) sentences INLINEFORM0 and one with (in-domain) pseudo-sentences INLINEFORM1 . The discriminator is a bidirectional-Recurrent Neural Network (RNN) of dimension 1024. Averaged states are passed to a single feed-forward layer, to which a sigmoid is applied. It inputs encodings of natural ( INLINEFORM2 ) and pseudo-sentences ( INLINEFORM3 ) and is trained to optimize:
INLINEFORM0
INLINEFORM0 's parameters are updated to maximally fool INLINEFORM1 , thus the loss INLINEFORM2 : INLINEFORM3
Finally, we keep the usual MT objective. ( INLINEFORM0 is a real or pseudo-sentence): INLINEFORM1
We thus need to train three sets of parameters: INLINEFORM0 and INLINEFORM1 (MT parameters), with INLINEFORM2 . The pseudo-source encoder and embeddings are updated wrt. both INLINEFORM3 and INLINEFORM4 . Following BIBREF28 , INLINEFORM5 is updated only when INLINEFORM6 's accuracy exceeds INLINEFORM7 . On the other hand, INLINEFORM8 is not updated when its accuracy exceeds INLINEFORM9 . At each update, two batches are generated for each type of data, which are encoded with the real or pseudo-encoder. The encoder outputs serve to compute INLINEFORM10 and INLINEFORM11 . Finally, the pseudo-source is encoded again (once INLINEFORM12 is updated), both encoders are plugged into the translation model and the MT cost is back-propagated down to the real and pseudo-word embeddings. Pseudo-encoder and discriminator parameters are pre-trained for 10k updates. At test time, the pseudo-encoder is ignored and inference is run as usual.
GANs can help
Results are in Table TABREF32 , assuming the same fine-tuning procedure as above. On top of the copy-marked setup, our GANs do not provide any improvement in both language pairs, with the exception of a small improvement for English-French on the out-of-domain test, which we understand as a sign that the model is more robust to domain variations, just like when adding pseudo-source noise. When combined with noise, the French model yields the best performance we could obtain with stupid BT on the in-domain tests, at least in terms of BLEU and BEER. On the News domain, we remain close to the baseline level, with slight improvements in German.
A first observation is that this method brings stupid BT models closer to conventional BT, at a greatly reduced computational cost. While French still remains 0.4 to 1.0 BLEU below very good backtranslation, both approaches are in the same ballpark for German - may be because BTs are better for the former system than for the latter.
Finally note that the GAN architecture has two differences with basic copy-marked: (a) a distinct encoder for real and pseudo-sentence; (b) a different training regime for these encoders. To sort out the effects of (a) and (b), we reproduce the GAN setup with BT sentences, instead of copies. Using a separate encoder for the pseudo-source in the backtrans-nmt setup can be detrimental to performance (see Table TABREF32 ): translation degrades in French for all metrics. Adding GANs on top of the pseudo-encoder was not able to make up for the degradation observed in French, but allowed the German system to slightly outperform backtrans-nmt. Even though this setup is unrealistic and overly costly, it shows that GANs are actually helping even good systems.
Using Target Language Models
In this section, we compare the previous methods with the use of a target side Language Model (LM). Several proposals exist in the literature to integrate LMs in NMT: for instance, BIBREF3 strengthen the decoder by integrating an extra, source independent, RNN layer in a conventional NMT architecture. Training is performed either with parallel, or monolingual data. In the latter case, word prediction only relies on the source independent part of the network.
LM Setup
We have followed BIBREF29 and reimplemented their deep-fusion technique. It requires to first independently learn a RNN-LM on the in-domain target data with a cross-entropy objective; then to train the optimal combination of the translation and the language models by adding the hidden state of the RNN-LM as an additional input to the softmax layer of the decoder. Our RNN-LMs are trained using dl4mt with the target side of the parallel data and the Europarl corpus (about 6M sentences for both French and German), using a one-layer GRU with the same dimension as the MT decoder (1024).
LM Results
Results are in Table TABREF33 . They show that deep-fusion hardly improves the Europarl results, while we obtain about +0.6 BLEU over the baseline on newstest-2014 for both languages. deep-fusion differs from stupid BT in that the model is not directly optimized on the in-domain data, but uses the LM trained on Europarl to maximize the likelihood of the out-of-domain training data. Therefore, no specific improvement is to be expected in terms of domain adaptation, and the performance increases in the more general domain. Combining deep-fusion and copy-marked + noise + GANs brings slight improvements on the German in-domain test sets, and performance out of the domain remains near the baseline level.
Re-analyzing the effects of BT
As a follow up of previous discussions, we analyze the effect of BT on the internals of the network. Arguably, using a copy of the target sentence instead of a natural source should not be helpful for the encoder, but is it also the case with a strong BT? What are the effects on the attention model?
Parameter freezing protocol
To investigate these questions, we run the same fine-tuning using the copy-marked, backtrans-nmt and backtrans-nmt setups. Note that except for the last one, all training scenarios have access to same target training data. We intend to see whether the overall performance of the NMT system degrades when we selectively freeze certain sets of parameters, meaning that they are not updated during fine-tuning.
Results
BLEU scores are in Table TABREF39 . The backtrans-nmt setup is hardly impacted by selective updates: updating the only decoder leads to a degradation of at most 0.2 BLEU. For copy-marked, we were not able to freeze the source embeddings, since these are initialized when fine-tuning begins and therefore need to be trained. We observe that freezing the encoder and/or the attention parameters has no impact on the English-German system, whereas it slightly degrades the English-French one. This suggests that using artificial sources, even of the poorest quality, has a positive impact on all the components of the network, which makes another big difference with the LM integration scenario.
The largest degradation is for natural, where the model is prevented from learning from informative source sentences, which leads to a decrease of 0.4 to over 1.0 BLEU. We assume from these experiments that BT impacts most of all the decoder, and learning to encode a pseudo-source, be it a copy or an actual back-translation, only marginally helps to significantly improve the quality. Finally, in the fwdtrans-nmt setup, freezing the decoder does not seem to harm learning with a natural source.
Related work
The literature devoted to the use of monolingual data is large, and quickly expanding. We already alluded to several possible ways to use such data: using back- or forward-translation or using a target language model. The former approach is mostly documented in BIBREF2 , and recently analyzed in BIBREF19 , which focus on fully artificial settings as well as pivot-based artificial data; and BIBREF4 , which studies the effects of increasing the size of BT data. The studies of BIBREF20 , BIBREF19 also consider forward translation and BIBREF21 expand these results to domain adaptation scenarios. Our results are complementary to these earlier studies.
As shown above, many alternatives to BT exist. The most obvious is to use target LMs BIBREF3 , BIBREF29 , as we have also done here; but attempts to improve the encoder using multi-task learning also exist BIBREF30 .
This investigation is also related to recent attempts to consider supplementary data with a valid target side, such as multi-lingual NMT BIBREF31 , where source texts in several languages are fed in the same encoder-decoder architecture, with partial sharing of the layers. This is another realistic scenario where additional resources can be used to selectively improve parts of the model.
Round trip training is another important source of inspiration, as it can be viewed as a way to use BT to perform semi-unsupervised BIBREF32 or unsupervised BIBREF33 training of NMT. The most convincing attempt to date along these lines has been proposed by BIBREF26 , who propose to use GANs to mitigate the difference between artificial and natural data.
Conclusion
In this paper, we have analyzed various ways to integrate monolingual data in an NMT framework, focusing on their impact on quality and domain adaptation. While confirming the effectiveness of BT, our study also proposed significantly cheaper ways to improve the baseline performance, using a slightly modified copy of the target, instead of its full BT. When no high quality BT is available, using GANs to make the pseudo-source sentences closer to natural source sentences is an efficient solution for domain adaptation.
To recap our answers to our initial questions: the quality of BT actually matters for NMT (cf. § SECREF8 ) and it seems that, even though artificial source are lexically less diverse and syntactically complex than real sentence, their monotonicity is a facilitating factor. We have studied cheaper alternatives and found out that copies of the target, if properly noised (§ SECREF4 ), and even better, if used with GANs, could be almost as good as low quality BTs (§ SECREF5 ): BT is only worth its cost when good BT can be generated. Finally, BT seems preferable to integrating external LM - at least in our data condition (§ SECREF6 ). Further experiments with larger LMs are needed to confirm this observation, and also to evaluate the complementarity of both strategies. More work is needed to better understand the impact of BT on subparts of the network (§ SECREF7 ).
In future work, we plan to investigate other cheap ways to generate artificial data. The experimental setup we proposed may also benefit from a refining of the data selection strategies to focus on the most useful monolingual sentences.
Question:
what data simulation techniques were introduced?
Answer:
| copy, copy-marked, copy-dummies | 328 | 26,501 | 26,503 | 26,561 | ... [The rest of the article is omitted]
|
79a44a68bb57b375d8a57a0a7f522d33476d9f33 | 79a44a68bb57b375d8a57a0a7f522d33476d9f33_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Knowledge and/or data is often modeled in a structure, such as indexes, tables, key-value pairs, or triplets. These data, by their nature (e.g., raw data or long time-series data), are not easily usable by humans; outlining their crucial need to be synthesized. Recently, numerous works have focused on leveraging structured data in various applications, such as question answering BIBREF0, BIBREF1 or table retrieval BIBREF2, BIBREF3. One emerging research field consists in transcribing data-structures into natural language in order to ease their understandablity and their usablity. This field is referred to as “data-to-text" BIBREF4 and has its place in several application domains (such as journalism BIBREF5 or medical diagnosis BIBREF6) or wide-audience applications (such as financial BIBREF7 and weather reports BIBREF8, or sport broadcasting BIBREF9, BIBREF10). As an example, Figure FIGREF1 shows a data-structure containing statistics on NBA basketball games, paired with its corresponding journalistic description.
Designing data-to-text models gives rise to two main challenges: 1) understanding structured data and 2) generating associated descriptions. Recent data-to-text models BIBREF11, BIBREF12, BIBREF13, BIBREF10 mostly rely on an encoder-decoder architecture BIBREF14 in which the data-structure is first encoded sequentially into a fixed-size vectorial representation by an encoder. Then, a decoder generates words conditioned on this representation. With the introduction of the attention mechanism BIBREF15 on one hand, which computes a context focused on important elements from the input at each decoding step and, on the other hand, the copy mechanism BIBREF16, BIBREF17 to deal with unknown or rare words, these systems produce fluent and domain comprehensive texts. For instance, Roberti et al. BIBREF18 train a character-wise encoder-decoder to generate descriptions of restaurants based on their attributes, while Puduppully et al. BIBREF12 design a more complex two-step decoder: they first generate a plan of elements to be mentioned, and then condition text generation on this plan. Although previous work yield overall good results, we identify two important caveats, that hinder precision (i.e. factual mentions) in the descriptions:
Linearization of the data-structure. In practice, most works focus on introducing innovating decoding modules, and still represent data as a unique sequence of elements to be encoded. For example, the table from Figure FIGREF1 would be linearized to [(Hawks, H/V, H), ..., (Magic, H/V, V), ...], effectively leading to losing distinction between rows, and therefore entities. To the best of our knowledge, only Liu et al. BIBREF19, BIBREF11 propose encoders constrained by the structure but these approaches are designed for single-entity structures.
Arbitrary ordering of unordered collections in recurrent networks (RNN). Most data-to-text systems use RNNs as encoders (such as GRUs or LSTMs), these architectures have however some limitations. Indeed, they require in practice their input to be fed sequentially. This way of encoding unordered sequences (i.e. collections of entities) implicitly assumes an arbitrary order within the collection which, as demonstrated by Vinyals et al. BIBREF20, significantly impacts the learning performance.
To address these shortcomings, we propose a new structured-data encoder assuming that structures should be hierarchically captured. Our contribution focuses on the encoding of the data-structure, thus the decoder is chosen to be a classical module as used in BIBREF12, BIBREF10. Our contribution is threefold:
We model the general structure of the data using a two-level architecture, first encoding all entities on the basis of their elements, then encoding the data structure on the basis of its entities;
We introduce the Transformer encoder BIBREF21 in data-to-text models to ensure robust encoding of each element/entities in comparison to all others, no matter their initial positioning;
We integrate a hierarchical attention mechanism to compute the hierarchical context fed into the decoder.
We report experiments on the RotoWire benchmark BIBREF10 which contains around $5K$ statistical tables of NBA basketball games paired with human-written descriptions. Our model is compared to several state-of-the-art models. Results show that the proposed architecture outperforms previous models on BLEU score and is generally better on qualitative metrics.
In the following, we first present a state-of-the art of data-to-text literature (Section 2), and then describe our proposed hierarchical data encoder (Section 3). The evaluation protocol is presented in Section 4, followed by the results (Section 5). Section 6 concludes the paper and presents perspectives.
Related Work
Until recently, efforts to bring out semantics from structured-data relied heavily on expert knowledge BIBREF22, BIBREF8. For example, in order to better transcribe numerical time series of weather data to a textual forecast, Reiter et al. BIBREF8 devise complex template schemes in collaboration with weather experts to build a consistent set of data-to-word rules.
Modern approaches to the wide range of tasks based on structured-data (e.g. table retrieval BIBREF2, BIBREF23, table classification BIBREF24, question answering BIBREF25) now propose to leverage progress in deep learning to represent these data into a semantic vector space (also called embedding space). In parallel, an emerging task, called “data-to-text", aims at describing structured data into a natural language description. This task stems from the neural machine translation (NMT) domain, and early work BIBREF26, BIBREF27, BIBREF10 represent the data records as a single sequence of facts to be entirely translated into natural language. Wiseman et al. BIBREF10 show the limits of traditional NMT systems on larger structured-data, where NMT systems fail to accurately extract salient elements.
To improve these models, a number of work BIBREF28, BIBREF12, BIBREF29 proposed innovating decoding modules based on planning and templates, to ensure factual and coherent mentions of records in generated descriptions. For example, Puduppully et al. BIBREF12 propose a two-step decoder which first targets specific records and then use them as a plan for the actual text generation. Similarly, Li et al. BIBREF28 proposed a delayed copy mechanism where their decoder also acts in two steps: 1) using a classical LSTM decoder to generate delexicalized text and 2) using a pointer network BIBREF30 to replace placeholders by records from the input data.
Closer to our work, very recent work BIBREF11, BIBREF19, BIBREF13 have proposed to take into account the data structure. More particularly, Puduppully et al. BIBREF13 follow entity-centric theories BIBREF31, BIBREF32 and propose a model based on dynamic entity representation at decoding time. It consists in conditioning the decoder on entity representations that are updated during inference at each decoding step. On the other hand, Liu et al. BIBREF11, BIBREF19 rather focus on introducing structure into the encoder. For instance, they propose a dual encoder BIBREF19 which encodes separately the sequence of element names and the sequence of element values. These approaches are however designed for single-entity data structures and do not account for delimitation between entities.
Our contribution differs from previous work in several aspects. First, instead of flatly concatenating elements from the data-structure and encoding them as a sequence BIBREF11, BIBREF12, BIBREF10, we constrain the encoding to the underlying structure of the input data, so that the delimitation between entities remains clear throughout the process. Second, unlike all works in the domain, we exploit the Transformer architecture BIBREF21 and leverage its particularity to directly compare elements with each others in order to avoid arbitrary assumptions on their ordering. Finally, in contrast to BIBREF33, BIBREF13 that use a complex updating mechanism to obtain a dynamic representation of the input data and its entities, we argue that explicit hierarchical encoding naturally guides the decoding process via hierarchical attention.
Hierarchical Encoder Model for Data-to-Text
In this section we introduce our proposed hierarchical model taking into account the data structure. We outline that the decoding component aiming to generate descriptions is considered as a black-box module so that our contribution is focused on the encoding module. We first describe the model overview, before detailing the hierarchical encoder and the associated hierarchical attention.
Hierarchical Encoder Model for Data-to-Text ::: Notation and General Overview
Let's consider the following notations:
$\bullet $ An entity $e_i$ is a set of $J_i$ unordered records $\lbrace r_{i,1}, ..., r_{i,j}, ..., r_{i,J_i}\rbrace $; where record $r_{i,j}$ is defined as a pair of key $k_{i,j}$ and value $v_{i,j}$. We outline that $J_i$ might differ between entities.
$\bullet $ A data-structure $s$ is an unordered set of $I$ entities $e_i$. We thus denote $s \lbrace e_1, ..., e_i, ..., e_I\rbrace $.
$\bullet $ For each data-structure, a textual description $y$ is associated. We refer to the first $t$ words of a description $y$ as $y_{1:t}$. Thus, the full sequence of words can be noted as $y = y_{1:T}$.
$\bullet $ The dataset $\mathcal {D}$ is a collection of $N$ aligned (data-structure, description) pairs $(s,y)$.
For instance, Figure FIGREF1 illustrates a data-structure associated with a description. The data-structure includes a set of entities (Hawks, Magic, Al Horford, Jeff Teague, ...). The entity Jeff Teague is modeled as a set of records {(PTS, 17), (REB, 0), (AST, 7) ...} in which, e.g., the record (PTS, 17) is characterized by a key (PTS) and a value (17).
For each data-structure $s$ in $\mathcal {D}$, the objective function aims to generate a description $\hat{y}$ as close as possible to the ground truth $y$. This objective function optimizes the following log-likelihood over the whole dataset $\mathcal {D}$:
where $\theta $ stands for the model parameters and $P(\hat{y}=y\ |\ s; \theta )$ the probability of the model to generate the adequate description $y$ for table $s$.
During inference, we generate the sequence $\hat{y}^*$ with the maximum a posteriori probability conditioned on table $s$. Using the chain rule, we get:
This equation is intractable in practice, we approximate a solution using beam search, as in BIBREF11, BIBREF19, BIBREF12, BIBREF13, BIBREF10.
Our model follows the encoder-decoder architecture BIBREF14. Because our contribution focuses on the encoding process, we chose the decoding module used in BIBREF12, BIBREF10: a two-layers LSTM network with a copy mechanism. In order to supervise this mechanism, we assume that each record value that also appears in the target is copied from the data-structure and we train the model to switch between freely generating words from the vocabulary and copying words from the input. We now describe the hierarchical encoder and the hierarchical attention.
Hierarchical Encoder Model for Data-to-Text ::: Hierarchical Encoding Model
As outlined in Section SECREF2, most previous work BIBREF28, BIBREF12, BIBREF13, BIBREF10, BIBREF29 make use of flat encoders that do not exploit the data structure. To keep the semantics of each element from the data-structure, we propose a hierarchical encoder which relies on two modules. The first one (module A in Figure FIGREF11) is called low-level encoder and encodes entities on the basis of their records; the second one (module B), called high-level encoder, encodes the data-structure on the basis of its underlying entities. In the low-level encoder, the traditional embedding layer is replaced by a record embedding layer as in BIBREF11, BIBREF12, BIBREF10. We present in what follows the record embedding layer and introduce our two hierarchical modules.
Hierarchical Encoder Model for Data-to-Text ::: Hierarchical Encoding Model ::: Record Embedding Layer.
The first layer of the network consists in learning two embedding matrices to embed the record keys and values. Keys $k_{i,j}$ are embedded to $\mathbf {k}_{i,j} \in \mathbb {R}^{d}$ and values $v_{i,j}$ to $\mathbf {v}_{i,j} \in \mathbb {R}^{d}$, with $d$ the size of the embedding. As in previous work BIBREF11, BIBREF12, BIBREF10, each record embedding $\mathbf {r}_{i,j}$ is computed by a linear projection on the concatenation $[\mathbf {k}_{i,j}$; $\mathbf {v}_{i,j}]$ followed by a non linearity:
where $\mathbf {W}_r \in \mathbb {R}^{2d \times d}$ and $\mathbf {b}_r \in \mathbb {R}^{d}$ are learnt parameters.
The low-level encoder aims at encoding a collection of records belonging to the same entity while the high-level encoder encodes the whole set of entities. Both the low-level and high-level encoders consider their input elements as unordered. We use the Transformer architecture from BIBREF21. For each encoder, we have the following peculiarities:
the Low-level encoder encodes each entity $e_i$ on the basis of its record embeddings $\mathbf {r}_{i,j}$. Each record embedding $\mathbf {r}_{i,j}$ is compared to other record embeddings to learn its final hidden representation $\mathbf {h}_{i,j}$. Furthermore, we add a special record [ENT] for each entity, illustrated in Figure FIGREF11 as the last record. Since entities might have a variable number of records, this token allows to aggregate final hidden record representations $\lbrace \mathbf {h}_{i,j}\rbrace _{j=1}^{J_i}$ in a fixed-sized representation vector $\mathbf {h}_{i}$.
the High-level encoder encodes the data-structure on the basis of its entity representation $\mathbf {h}_{i}$. Similarly to the Low-level encoder, the final hidden state $\mathbf {e_i}$ of an entity is computed by comparing entity representation $\mathbf {h}_{i}$ with each others. The data-structure representation $\mathbf {z}$ is computed as the mean of these entity representations, and is used for the decoder initialization.
Hierarchical Encoder Model for Data-to-Text ::: Hierarchical attention
To fully leverage the hierarchical structure of our encoder, we propose two variants of hierarchical attention mechanism to compute the context fed to the decoder module.
$\bullet $ Traditional Hierarchical Attention. As in BIBREF13, we hypothesize that a dynamic context should be computed in two steps: first attending to entities, then to records corresponding to these entities. To implement this hierarchical attention, at each decoding step $t$, the model learns a first set of attention scores $\alpha _{i,t}$ over entities $e_i$ and a second set of attention scores $\beta _{i,j,t}$ over records $r_{i,j}$ belonging to entity $e_i$. The $\alpha _{i,t}$ scores are normalized to form a distribution over all entities $e_i$, and $\beta _{i,j,t}$ scores are normalized to form a distribution over records $r_{i,j}$ of entity $e_i$. Each entity is then represented as a weighted sum of its record embeddings, and the entire data structure is represented as a weighted sum of the entity representations. The dynamic context is computed as:
ct = i=1I (i,t ( j i,j,t ri,j ))
where i,t exp(dtWei) and i,j,t exp(dtWhi,j)
where $\mathbf {d_t}$ is the decoder hidden state at time step $t$, $\mathbf {W}_{\alpha } \in \mathbb {R}^{d\times d}$ and $\mathbf {W}_{\beta } \in \mathbb {R}^{d\times d}$ are learnt parameters, $ \sum _i\alpha _{i,t} = 1$, and for all $i \in \lbrace 1,...,I\rbrace $ $\sum _{j}\beta _{i,j,t} = 1$.
$\bullet $ Key-guided Hierarchical Attention. This variant follows the intuition that once an entity is chosen for mention (thanks to $\alpha _{i,t}$), only the type of records is important to determine the content of the description. For example, when deciding to mention a player, all experts automatically report his score without consideration of its specific value. To test this intuition, we model the attention scores by computing the $\beta _{i,j,t}$ scores from equation (SECREF16) solely on the embedding of the key rather than on the full record representation $\mathbf {h}_{i,j}$:
Please note that the different embeddings and the model parameters presented in the model components are learnt using Equation 1.
Experimental setup ::: The Rotowire dataset
To evaluate the effectiveness of our model, and demonstrate its flexibility at handling heavy data-structure made of several types of entities, we used the RotoWire dataset BIBREF10. It includes basketball games statistical tables paired with journalistic descriptions of the games, as can be seen in the example of Figure FIGREF1. The descriptions are professionally written and average 337 words with a vocabulary size of $11.3$K. There are 39 different record keys, and the average number of records (resp. entities) in a single data-structure is 628 (resp. 28). Entities are of two types, either team or player, and player descriptions depend on their involvement in the game. We followed the data partitions introduced with the dataset and used a train/validation/test sets of respectively $3,398$/727/728 (data-structure, description) pairs.
Experimental setup ::: Evaluation metrics
We evaluate our model through two types of metrics. The BLEU score BIBREF34 aims at measuring to what extent the generated descriptions are literally closed to the ground truth. The second category designed by BIBREF10 is more qualitative.
Experimental setup ::: Evaluation metrics ::: BLEU Score.
The BLEU score BIBREF34 is commonly used as an evaluation metric in text generation tasks. It estimates the correspondence between a machine output and that of a human by computing the number of co-occurrences for ngrams ($n \in {1, 2, 3, 4}$) between the generated candidate and the ground truth. We use the implementation code released by BIBREF35.
Experimental setup ::: Evaluation metrics ::: Information extraction-oriented metrics.
These metrics estimate the ability of our model to integrate elements from the table in its descriptions. Particularly, they compare the gold and generated descriptions and measure to what extent the extracted relations are aligned or differ. To do so, we follow the protocol presented in BIBREF10. First, we apply an information extraction (IE) system trained on labeled relations from the gold descriptions of the RotoWire train dataset. Entity-value pairs are extracted from the descriptions. For example, in the sentence Isaiah Thomas led the team in scoring, totaling 23 points [...]., an IE tool will extract the pair (Isaiah Thomas, 23, PTS). Second, we compute three metrics on the extracted information:
$\bullet $ Relation Generation (RG) estimates how well the system is able to generate text containing factual (i.e., correct) records. We measure the precision and absolute number (denoted respectively RG-P% and RG-#) of unique relations $r$ extracted from $\hat{y}_{1:T}$ that also appear in $s$.
$\bullet $ Content Selection (CS) measures how well the generated document matches the gold document in terms of mentioned records. We measure the precision and recall (denoted respectively CS-P% and CS-R%) of unique relations $r$ extracted from $\hat{y}_{1:T}$ that are also extracted from $y_{1:T}$.
$\bullet $ Content Ordering (CO) analyzes how well the system orders the records discussed in the description. We measure the normalized Damerau-Levenshtein distance BIBREF36 between the sequences of records extracted from $\hat{y}_{1:T}$ that are also extracted from $y_{1:T}$.
CS primarily targets the “what to say" aspect of evaluation, CO targets the “how to say it" aspect, and RG targets both. Note that for CS, CO, RG-% and BLEU metrics, higher is better; which is not true for RG-#. The IE system used in the experiments is able to extract an average of 17 factual records from gold descriptions. In order to mimic a human expert, a generative system should approach this number and not overload generation with brute facts.
Experimental setup ::: Baselines
We compare our hierarchical model against three systems. For each of them, we report the results of the best performing models presented in each paper.
$\bullet $ Wiseman BIBREF10 is a standard encoder-decoder system with copy mechanism.
$\bullet $ Li BIBREF28 is a standard encoder-decoder with a delayed copy mechanism: text is first generated with placeholders, which are replaced by salient records extracted from the table by a pointer network.
$\bullet $ Puduppully-plan BIBREF12 acts in two steps: a first standard encoder-decoder generates a plan, i.e. a list of salient records from the table; a second standard encoder-decoder generates text from this plan.
$\bullet $ Puduppully-updt BIBREF13. It consists in a standard encoder-decoder, with an added module aimed at updating record representations during the generation process. At each decoding step, a gated recurrent network computes which records should be updated and what should be their new representation.
Experimental setup ::: Baselines ::: Model scenarios
We test the importance of the input structure by training different variants of the proposed architecture:
$\bullet $ Flat, where we feed the input sequentially to the encoder, losing all notion of hierarchy. As a consequence, the model uses standard attention. This variant is closest to Wiseman, with the exception that we use a Transformer to encode the input sequence instead of an RNN.
$\bullet $ Hierarchical-kv is our full hierarchical model, with traditional hierarchical attention, i.e. where attention over records is computed on the full record encoding, as in equation (SECREF16).
$\bullet $ Hierarchical-k is our full hierarchical model, with key-guided hierarchical attention, i.e. where attention over records is computed only on the record key representations, as in equation (DISPLAY_FORM17).
Experimental setup ::: Implementation details
The decoder is the one used in BIBREF12, BIBREF13, BIBREF10 with the same hyper-parameters. For the encoder module, both the low-level and high-level encoders use a two-layers multi-head self-attention with two heads. To fit with the small number of record keys in our dataset (39), their embedding size is fixed to 20. The size of the record value embeddings and hidden layers of the Transformer encoders are both set to 300. We use dropout at rate 0.5. The models are trained with a batch size of 64. We follow the training procedure in BIBREF21 and train the model for a fixed number of 25K updates, and average the weights of the last 5 checkpoints (at every 1K updates) to ensure more stability across runs. All models were trained with the Adam optimizer BIBREF37; the initial learning rate is 0.001, and is reduced by half every 10K steps. We used beam search with beam size of 5 during inference. All the models are implemented in OpenNMT-py BIBREF38. All code is available at https://github.com/KaijuML/data-to-text-hierarchical
Results
Our results on the RotoWire testset are summarized in Table TABREF25. For each proposed variant of our architecture, we report the mean score over ten runs, as well as the standard deviation in subscript. Results are compared to baselines BIBREF12, BIBREF13, BIBREF10 and variants of our models. We also report the result of the oracle (metrics on the gold descriptions). Please note that gold descriptions trivially obtain 100% on all metrics expect RG, as they are all based on comparison with themselves. RG scores are different, as the IE system is imperfect and fails to extract accurate entities 4% of the time. RG-# is an absolute count.
Results ::: Ablation studies
To evaluate the impact of our model components, we first compare scenarios Flat, Hierarchical-k, and Hierarchical-kv. As shown in Table TABREF25, we can see the lower results obtained by the Flat scenario compared to the other scenarios (e.g. BLEU $16.7$ vs. $17.5$ for resp. Flat and Hierarchical-k), suggesting the effectiveness of encoding the data-structure using a hierarchy. This is expected, as losing explicit delimitation between entities makes it harder a) for the encoder to encode semantics of the objects contained in the table and b) for the attention mechanism to extract salient entities/records.
Second, the comparison between scenario Hierarchical-kv and Hierarchical-k shows that omitting entirely the influence of the record values in the attention mechanism is more effective: this last variant performs slightly better in all metrics excepted CS-R%, reinforcing our intuition that focusing on the structure modeling is an important part of data encoding as well as confirming the intuition explained in Section SECREF16: once an entity is selected, facts about this entity are relevant based on their key, not value which might add noise. To illustrate this intuition, we depict in Figure FIGREF27 attention scores (recall $\alpha _{i,t}$ and $\beta _{i,j,t}$ from equations (SECREF16) and (DISPLAY_FORM17)) for both variants Hierarchical-kv and Hierarchical-k. We particularly focus on the timestamp where the models should mention the number of points scored during the first quarter of the game. Scores of Hierarchical-k are sharp, with all of the weight on the correct record (PTS_QTR1, 26) whereas scores of Hierarchical-kv are more distributed over all PTS_QTR records, ultimately failing to retrieve the correct one.
Results ::: Comparison w.r.t. baselines.
From a general point of view, we can see from Table TABREF25 that our scenarios obtain significantly higher results in terms of BLEU over all models; our best model Hierarchical-k reaching $17.5$ vs. $16.5$ against the best baseline. This means that our models learns to generate fluent sequences of words, close to the gold descriptions, adequately picking up on domain lingo. Qualitative metrics are either better or on par with baselines. We show in Figure FIGREF29 a text generated by our best model, which can be directly compared to the gold description in Figure FIGREF1. Generation is fluent and contains domain-specific expressions. As reflected in Table TABREF25, the number of correct mentions (in green) outweights the number of incorrect mentions (in red). Please note that, as in previous work BIBREF28, BIBREF12, BIBREF13, BIBREF10, generated texts still contain a number of incorrect facts, as well hallucinations (in blue): sentences that have no basis in the input data (e.g. “[...] he's now averaging 22 points [...]."). While not the direct focus of our work, this highlights that any operation meant to enrich the semantics of structured data can also enrich the data with incorrect facts.
Specifically, regarding all baselines, we can outline the following statements.
$\bullet $ Our hierarchical models achieve significantly better scores on all metrics when compared to the flat architecture Wiseman, reinforcing the crucial role of structure in data semantics and saliency. The analysis of RG metrics shows that Wiseman seems to be the more naturalistic in terms of number of factual mentions (RG#) since it is the closest scenario to the gold value (16.83 vs. 17.31 for resp. Wiseman and Hierarchical-k). However, Wiseman achieves only $75.62$% of precision, effectively mentioning on average a total of $22.25$ records (wrong or accurate), where our model Hierarchical-k scores a precision of $89.46$%, leading to $23.66$ total mentions, just slightly above Wiseman.
$\bullet $ The comparison between the Flat scenario and Wiseman is particularly interesting. Indeed, these two models share the same intuition to flatten the data-structure. The only difference stands on the encoder mechanism: bi-LSTM vs. Transformer, for Wiseman and Flat respectively. Results shows that our Flat scenario obtains a significant higher BLEU score (16.7 vs. 14.5) and generates fluent descriptions with accurate mentions (RG-P%) that are also included in the gold descriptions (CS-R%). This suggests that introducing the Transformer architecture is promising way to implicitly account for data structure.
$\bullet $ Our hierarchical models outperform the two-step decoders of Li and Puduppully-plan on both BLEU and all qualitative metrics, showing that capturing structure in the encoding process is more effective that predicting a structure in the decoder (i.e., planning or templating). While our models sensibly outperform in precision at factual mentions, the baseline Puduppully-plan reaches $34.28$ mentions on average, showing that incorporating modules dedicated to entity extraction leads to over-focusing on entities; contrasting with our models that learn to generate more balanced descriptions.
$\bullet $ The comparison with Puduppully-updt shows that dynamically updating the encoding across the generation process can lead to better Content Ordering (CO) and RG-P%. However, this does not help with Content Selection (CS) since our best model Hierarchical-k obtains slightly better scores. Indeed, Puduppully-updt updates representations after each mention allowing to keep track of the mention history. This guides the ordering of mentions (CO metric), each step limiting more the number of candidate mentions (increasing RG-P%). In contrast, our model encodes saliency among records/entities more effectively (CS metric). We note that while our model encodes the data-structure once and for all, Puduppully-updt recomputes, via the updates, the encoding at each step and therefore significantly increases computation complexity. Combined with their RG-# score of $30.11$, we argue that our model is simpler, and obtains fluent description with accurate mentions in a more human-like fashion.
We would also like to draw attention to the number of parameters used by those architectures. We note that our scenarios relies on a lower number of parameters (14 millions) compared to all baselines (ranging from 23 to 45 millions). This outlines the effectiveness in the design of our model relying on a structure encoding, in contrast to other approach that try to learn the structure of data/descriptions from a linearized encoding.
Conclusion and future work
In this work we have proposed a hierarchical encoder for structured data, which 1) leverages the structure to form efficient representation of its input; 2) has strong synergy with the hierarchical attention of its associated decoder. This results in an effective and more light-weight model. Experimental evaluation on the RotoWire benchmark shows that our model outperforms competitive baselines in terms of BLEU score and is generally better on qualitative metrics. This way of representing structured databases may lead to automatic inference and enrichment, e.g., by comparing entities. This direction could be driven by very recent operation-guided networks BIBREF39, BIBREF40. In addition, we note that our approach can still lead to erroneous facts or even hallucinations. An interesting perspective might be to further constrain the model on the data structure in order to prevent inaccurate of even contradictory descriptions.
Acknowledgements
We would like to thank the H2020 project AI4EU (825619) which partially supports Laure Soulier and Patrick Gallinari.
Question:
Which qualitative metric are used for evaluation?
Answer:
| Relation Generation (RG) , Content Selection (CS), Content Ordering (CO) | 328 | 31,546 | 31,548 | 31,607 | ... [The rest of the article is omitted]
|
79a44a68bb57b375d8a57a0a7f522d33476d9f33 | 79a44a68bb57b375d8a57a0a7f522d33476d9f33_1 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Knowledge and/or data is often modeled in a structure, such as indexes, tables, key-value pairs, or triplets. These data, by their nature (e.g., raw data or long time-series data), are not easily usable by humans; outlining their crucial need to be synthesized. Recently, numerous works have focused on leveraging structured data in various applications, such as question answering BIBREF0, BIBREF1 or table retrieval BIBREF2, BIBREF3. One emerging research field consists in transcribing data-structures into natural language in order to ease their understandablity and their usablity. This field is referred to as “data-to-text" BIBREF4 and has its place in several application domains (such as journalism BIBREF5 or medical diagnosis BIBREF6) or wide-audience applications (such as financial BIBREF7 and weather reports BIBREF8, or sport broadcasting BIBREF9, BIBREF10). As an example, Figure FIGREF1 shows a data-structure containing statistics on NBA basketball games, paired with its corresponding journalistic description.
Designing data-to-text models gives rise to two main challenges: 1) understanding structured data and 2) generating associated descriptions. Recent data-to-text models BIBREF11, BIBREF12, BIBREF13, BIBREF10 mostly rely on an encoder-decoder architecture BIBREF14 in which the data-structure is first encoded sequentially into a fixed-size vectorial representation by an encoder. Then, a decoder generates words conditioned on this representation. With the introduction of the attention mechanism BIBREF15 on one hand, which computes a context focused on important elements from the input at each decoding step and, on the other hand, the copy mechanism BIBREF16, BIBREF17 to deal with unknown or rare words, these systems produce fluent and domain comprehensive texts. For instance, Roberti et al. BIBREF18 train a character-wise encoder-decoder to generate descriptions of restaurants based on their attributes, while Puduppully et al. BIBREF12 design a more complex two-step decoder: they first generate a plan of elements to be mentioned, and then condition text generation on this plan. Although previous work yield overall good results, we identify two important caveats, that hinder precision (i.e. factual mentions) in the descriptions:
Linearization of the data-structure. In practice, most works focus on introducing innovating decoding modules, and still represent data as a unique sequence of elements to be encoded. For example, the table from Figure FIGREF1 would be linearized to [(Hawks, H/V, H), ..., (Magic, H/V, V), ...], effectively leading to losing distinction between rows, and therefore entities. To the best of our knowledge, only Liu et al. BIBREF19, BIBREF11 propose encoders constrained by the structure but these approaches are designed for single-entity structures.
Arbitrary ordering of unordered collections in recurrent networks (RNN). Most data-to-text systems use RNNs as encoders (such as GRUs or LSTMs), these architectures have however some limitations. Indeed, they require in practice their input to be fed sequentially. This way of encoding unordered sequences (i.e. collections of entities) implicitly assumes an arbitrary order within the collection which, as demonstrated by Vinyals et al. BIBREF20, significantly impacts the learning performance.
To address these shortcomings, we propose a new structured-data encoder assuming that structures should be hierarchically captured. Our contribution focuses on the encoding of the data-structure, thus the decoder is chosen to be a classical module as used in BIBREF12, BIBREF10. Our contribution is threefold:
We model the general structure of the data using a two-level architecture, first encoding all entities on the basis of their elements, then encoding the data structure on the basis of its entities;
We introduce the Transformer encoder BIBREF21 in data-to-text models to ensure robust encoding of each element/entities in comparison to all others, no matter their initial positioning;
We integrate a hierarchical attention mechanism to compute the hierarchical context fed into the decoder.
We report experiments on the RotoWire benchmark BIBREF10 which contains around $5K$ statistical tables of NBA basketball games paired with human-written descriptions. Our model is compared to several state-of-the-art models. Results show that the proposed architecture outperforms previous models on BLEU score and is generally better on qualitative metrics.
In the following, we first present a state-of-the art of data-to-text literature (Section 2), and then describe our proposed hierarchical data encoder (Section 3). The evaluation protocol is presented in Section 4, followed by the results (Section 5). Section 6 concludes the paper and presents perspectives.
Related Work
Until recently, efforts to bring out semantics from structured-data relied heavily on expert knowledge BIBREF22, BIBREF8. For example, in order to better transcribe numerical time series of weather data to a textual forecast, Reiter et al. BIBREF8 devise complex template schemes in collaboration with weather experts to build a consistent set of data-to-word rules.
Modern approaches to the wide range of tasks based on structured-data (e.g. table retrieval BIBREF2, BIBREF23, table classification BIBREF24, question answering BIBREF25) now propose to leverage progress in deep learning to represent these data into a semantic vector space (also called embedding space). In parallel, an emerging task, called “data-to-text", aims at describing structured data into a natural language description. This task stems from the neural machine translation (NMT) domain, and early work BIBREF26, BIBREF27, BIBREF10 represent the data records as a single sequence of facts to be entirely translated into natural language. Wiseman et al. BIBREF10 show the limits of traditional NMT systems on larger structured-data, where NMT systems fail to accurately extract salient elements.
To improve these models, a number of work BIBREF28, BIBREF12, BIBREF29 proposed innovating decoding modules based on planning and templates, to ensure factual and coherent mentions of records in generated descriptions. For example, Puduppully et al. BIBREF12 propose a two-step decoder which first targets specific records and then use them as a plan for the actual text generation. Similarly, Li et al. BIBREF28 proposed a delayed copy mechanism where their decoder also acts in two steps: 1) using a classical LSTM decoder to generate delexicalized text and 2) using a pointer network BIBREF30 to replace placeholders by records from the input data.
Closer to our work, very recent work BIBREF11, BIBREF19, BIBREF13 have proposed to take into account the data structure. More particularly, Puduppully et al. BIBREF13 follow entity-centric theories BIBREF31, BIBREF32 and propose a model based on dynamic entity representation at decoding time. It consists in conditioning the decoder on entity representations that are updated during inference at each decoding step. On the other hand, Liu et al. BIBREF11, BIBREF19 rather focus on introducing structure into the encoder. For instance, they propose a dual encoder BIBREF19 which encodes separately the sequence of element names and the sequence of element values. These approaches are however designed for single-entity data structures and do not account for delimitation between entities.
Our contribution differs from previous work in several aspects. First, instead of flatly concatenating elements from the data-structure and encoding them as a sequence BIBREF11, BIBREF12, BIBREF10, we constrain the encoding to the underlying structure of the input data, so that the delimitation between entities remains clear throughout the process. Second, unlike all works in the domain, we exploit the Transformer architecture BIBREF21 and leverage its particularity to directly compare elements with each others in order to avoid arbitrary assumptions on their ordering. Finally, in contrast to BIBREF33, BIBREF13 that use a complex updating mechanism to obtain a dynamic representation of the input data and its entities, we argue that explicit hierarchical encoding naturally guides the decoding process via hierarchical attention.
Hierarchical Encoder Model for Data-to-Text
In this section we introduce our proposed hierarchical model taking into account the data structure. We outline that the decoding component aiming to generate descriptions is considered as a black-box module so that our contribution is focused on the encoding module. We first describe the model overview, before detailing the hierarchical encoder and the associated hierarchical attention.
Hierarchical Encoder Model for Data-to-Text ::: Notation and General Overview
Let's consider the following notations:
$\bullet $ An entity $e_i$ is a set of $J_i$ unordered records $\lbrace r_{i,1}, ..., r_{i,j}, ..., r_{i,J_i}\rbrace $; where record $r_{i,j}$ is defined as a pair of key $k_{i,j}$ and value $v_{i,j}$. We outline that $J_i$ might differ between entities.
$\bullet $ A data-structure $s$ is an unordered set of $I$ entities $e_i$. We thus denote $s \lbrace e_1, ..., e_i, ..., e_I\rbrace $.
$\bullet $ For each data-structure, a textual description $y$ is associated. We refer to the first $t$ words of a description $y$ as $y_{1:t}$. Thus, the full sequence of words can be noted as $y = y_{1:T}$.
$\bullet $ The dataset $\mathcal {D}$ is a collection of $N$ aligned (data-structure, description) pairs $(s,y)$.
For instance, Figure FIGREF1 illustrates a data-structure associated with a description. The data-structure includes a set of entities (Hawks, Magic, Al Horford, Jeff Teague, ...). The entity Jeff Teague is modeled as a set of records {(PTS, 17), (REB, 0), (AST, 7) ...} in which, e.g., the record (PTS, 17) is characterized by a key (PTS) and a value (17).
For each data-structure $s$ in $\mathcal {D}$, the objective function aims to generate a description $\hat{y}$ as close as possible to the ground truth $y$. This objective function optimizes the following log-likelihood over the whole dataset $\mathcal {D}$:
where $\theta $ stands for the model parameters and $P(\hat{y}=y\ |\ s; \theta )$ the probability of the model to generate the adequate description $y$ for table $s$.
During inference, we generate the sequence $\hat{y}^*$ with the maximum a posteriori probability conditioned on table $s$. Using the chain rule, we get:
This equation is intractable in practice, we approximate a solution using beam search, as in BIBREF11, BIBREF19, BIBREF12, BIBREF13, BIBREF10.
Our model follows the encoder-decoder architecture BIBREF14. Because our contribution focuses on the encoding process, we chose the decoding module used in BIBREF12, BIBREF10: a two-layers LSTM network with a copy mechanism. In order to supervise this mechanism, we assume that each record value that also appears in the target is copied from the data-structure and we train the model to switch between freely generating words from the vocabulary and copying words from the input. We now describe the hierarchical encoder and the hierarchical attention.
Hierarchical Encoder Model for Data-to-Text ::: Hierarchical Encoding Model
As outlined in Section SECREF2, most previous work BIBREF28, BIBREF12, BIBREF13, BIBREF10, BIBREF29 make use of flat encoders that do not exploit the data structure. To keep the semantics of each element from the data-structure, we propose a hierarchical encoder which relies on two modules. The first one (module A in Figure FIGREF11) is called low-level encoder and encodes entities on the basis of their records; the second one (module B), called high-level encoder, encodes the data-structure on the basis of its underlying entities. In the low-level encoder, the traditional embedding layer is replaced by a record embedding layer as in BIBREF11, BIBREF12, BIBREF10. We present in what follows the record embedding layer and introduce our two hierarchical modules.
Hierarchical Encoder Model for Data-to-Text ::: Hierarchical Encoding Model ::: Record Embedding Layer.
The first layer of the network consists in learning two embedding matrices to embed the record keys and values. Keys $k_{i,j}$ are embedded to $\mathbf {k}_{i,j} \in \mathbb {R}^{d}$ and values $v_{i,j}$ to $\mathbf {v}_{i,j} \in \mathbb {R}^{d}$, with $d$ the size of the embedding. As in previous work BIBREF11, BIBREF12, BIBREF10, each record embedding $\mathbf {r}_{i,j}$ is computed by a linear projection on the concatenation $[\mathbf {k}_{i,j}$; $\mathbf {v}_{i,j}]$ followed by a non linearity:
where $\mathbf {W}_r \in \mathbb {R}^{2d \times d}$ and $\mathbf {b}_r \in \mathbb {R}^{d}$ are learnt parameters.
The low-level encoder aims at encoding a collection of records belonging to the same entity while the high-level encoder encodes the whole set of entities. Both the low-level and high-level encoders consider their input elements as unordered. We use the Transformer architecture from BIBREF21. For each encoder, we have the following peculiarities:
the Low-level encoder encodes each entity $e_i$ on the basis of its record embeddings $\mathbf {r}_{i,j}$. Each record embedding $\mathbf {r}_{i,j}$ is compared to other record embeddings to learn its final hidden representation $\mathbf {h}_{i,j}$. Furthermore, we add a special record [ENT] for each entity, illustrated in Figure FIGREF11 as the last record. Since entities might have a variable number of records, this token allows to aggregate final hidden record representations $\lbrace \mathbf {h}_{i,j}\rbrace _{j=1}^{J_i}$ in a fixed-sized representation vector $\mathbf {h}_{i}$.
the High-level encoder encodes the data-structure on the basis of its entity representation $\mathbf {h}_{i}$. Similarly to the Low-level encoder, the final hidden state $\mathbf {e_i}$ of an entity is computed by comparing entity representation $\mathbf {h}_{i}$ with each others. The data-structure representation $\mathbf {z}$ is computed as the mean of these entity representations, and is used for the decoder initialization.
Hierarchical Encoder Model for Data-to-Text ::: Hierarchical attention
To fully leverage the hierarchical structure of our encoder, we propose two variants of hierarchical attention mechanism to compute the context fed to the decoder module.
$\bullet $ Traditional Hierarchical Attention. As in BIBREF13, we hypothesize that a dynamic context should be computed in two steps: first attending to entities, then to records corresponding to these entities. To implement this hierarchical attention, at each decoding step $t$, the model learns a first set of attention scores $\alpha _{i,t}$ over entities $e_i$ and a second set of attention scores $\beta _{i,j,t}$ over records $r_{i,j}$ belonging to entity $e_i$. The $\alpha _{i,t}$ scores are normalized to form a distribution over all entities $e_i$, and $\beta _{i,j,t}$ scores are normalized to form a distribution over records $r_{i,j}$ of entity $e_i$. Each entity is then represented as a weighted sum of its record embeddings, and the entire data structure is represented as a weighted sum of the entity representations. The dynamic context is computed as:
ct = i=1I (i,t ( j i,j,t ri,j ))
where i,t exp(dtWei) and i,j,t exp(dtWhi,j)
where $\mathbf {d_t}$ is the decoder hidden state at time step $t$, $\mathbf {W}_{\alpha } \in \mathbb {R}^{d\times d}$ and $\mathbf {W}_{\beta } \in \mathbb {R}^{d\times d}$ are learnt parameters, $ \sum _i\alpha _{i,t} = 1$, and for all $i \in \lbrace 1,...,I\rbrace $ $\sum _{j}\beta _{i,j,t} = 1$.
$\bullet $ Key-guided Hierarchical Attention. This variant follows the intuition that once an entity is chosen for mention (thanks to $\alpha _{i,t}$), only the type of records is important to determine the content of the description. For example, when deciding to mention a player, all experts automatically report his score without consideration of its specific value. To test this intuition, we model the attention scores by computing the $\beta _{i,j,t}$ scores from equation (SECREF16) solely on the embedding of the key rather than on the full record representation $\mathbf {h}_{i,j}$:
Please note that the different embeddings and the model parameters presented in the model components are learnt using Equation 1.
Experimental setup ::: The Rotowire dataset
To evaluate the effectiveness of our model, and demonstrate its flexibility at handling heavy data-structure made of several types of entities, we used the RotoWire dataset BIBREF10. It includes basketball games statistical tables paired with journalistic descriptions of the games, as can be seen in the example of Figure FIGREF1. The descriptions are professionally written and average 337 words with a vocabulary size of $11.3$K. There are 39 different record keys, and the average number of records (resp. entities) in a single data-structure is 628 (resp. 28). Entities are of two types, either team or player, and player descriptions depend on their involvement in the game. We followed the data partitions introduced with the dataset and used a train/validation/test sets of respectively $3,398$/727/728 (data-structure, description) pairs.
Experimental setup ::: Evaluation metrics
We evaluate our model through two types of metrics. The BLEU score BIBREF34 aims at measuring to what extent the generated descriptions are literally closed to the ground truth. The second category designed by BIBREF10 is more qualitative.
Experimental setup ::: Evaluation metrics ::: BLEU Score.
The BLEU score BIBREF34 is commonly used as an evaluation metric in text generation tasks. It estimates the correspondence between a machine output and that of a human by computing the number of co-occurrences for ngrams ($n \in {1, 2, 3, 4}$) between the generated candidate and the ground truth. We use the implementation code released by BIBREF35.
Experimental setup ::: Evaluation metrics ::: Information extraction-oriented metrics.
These metrics estimate the ability of our model to integrate elements from the table in its descriptions. Particularly, they compare the gold and generated descriptions and measure to what extent the extracted relations are aligned or differ. To do so, we follow the protocol presented in BIBREF10. First, we apply an information extraction (IE) system trained on labeled relations from the gold descriptions of the RotoWire train dataset. Entity-value pairs are extracted from the descriptions. For example, in the sentence Isaiah Thomas led the team in scoring, totaling 23 points [...]., an IE tool will extract the pair (Isaiah Thomas, 23, PTS). Second, we compute three metrics on the extracted information:
$\bullet $ Relation Generation (RG) estimates how well the system is able to generate text containing factual (i.e., correct) records. We measure the precision and absolute number (denoted respectively RG-P% and RG-#) of unique relations $r$ extracted from $\hat{y}_{1:T}$ that also appear in $s$.
$\bullet $ Content Selection (CS) measures how well the generated document matches the gold document in terms of mentioned records. We measure the precision and recall (denoted respectively CS-P% and CS-R%) of unique relations $r$ extracted from $\hat{y}_{1:T}$ that are also extracted from $y_{1:T}$.
$\bullet $ Content Ordering (CO) analyzes how well the system orders the records discussed in the description. We measure the normalized Damerau-Levenshtein distance BIBREF36 between the sequences of records extracted from $\hat{y}_{1:T}$ that are also extracted from $y_{1:T}$.
CS primarily targets the “what to say" aspect of evaluation, CO targets the “how to say it" aspect, and RG targets both. Note that for CS, CO, RG-% and BLEU metrics, higher is better; which is not true for RG-#. The IE system used in the experiments is able to extract an average of 17 factual records from gold descriptions. In order to mimic a human expert, a generative system should approach this number and not overload generation with brute facts.
Experimental setup ::: Baselines
We compare our hierarchical model against three systems. For each of them, we report the results of the best performing models presented in each paper.
$\bullet $ Wiseman BIBREF10 is a standard encoder-decoder system with copy mechanism.
$\bullet $ Li BIBREF28 is a standard encoder-decoder with a delayed copy mechanism: text is first generated with placeholders, which are replaced by salient records extracted from the table by a pointer network.
$\bullet $ Puduppully-plan BIBREF12 acts in two steps: a first standard encoder-decoder generates a plan, i.e. a list of salient records from the table; a second standard encoder-decoder generates text from this plan.
$\bullet $ Puduppully-updt BIBREF13. It consists in a standard encoder-decoder, with an added module aimed at updating record representations during the generation process. At each decoding step, a gated recurrent network computes which records should be updated and what should be their new representation.
Experimental setup ::: Baselines ::: Model scenarios
We test the importance of the input structure by training different variants of the proposed architecture:
$\bullet $ Flat, where we feed the input sequentially to the encoder, losing all notion of hierarchy. As a consequence, the model uses standard attention. This variant is closest to Wiseman, with the exception that we use a Transformer to encode the input sequence instead of an RNN.
$\bullet $ Hierarchical-kv is our full hierarchical model, with traditional hierarchical attention, i.e. where attention over records is computed on the full record encoding, as in equation (SECREF16).
$\bullet $ Hierarchical-k is our full hierarchical model, with key-guided hierarchical attention, i.e. where attention over records is computed only on the record key representations, as in equation (DISPLAY_FORM17).
Experimental setup ::: Implementation details
The decoder is the one used in BIBREF12, BIBREF13, BIBREF10 with the same hyper-parameters. For the encoder module, both the low-level and high-level encoders use a two-layers multi-head self-attention with two heads. To fit with the small number of record keys in our dataset (39), their embedding size is fixed to 20. The size of the record value embeddings and hidden layers of the Transformer encoders are both set to 300. We use dropout at rate 0.5. The models are trained with a batch size of 64. We follow the training procedure in BIBREF21 and train the model for a fixed number of 25K updates, and average the weights of the last 5 checkpoints (at every 1K updates) to ensure more stability across runs. All models were trained with the Adam optimizer BIBREF37; the initial learning rate is 0.001, and is reduced by half every 10K steps. We used beam search with beam size of 5 during inference. All the models are implemented in OpenNMT-py BIBREF38. All code is available at https://github.com/KaijuML/data-to-text-hierarchical
Results
Our results on the RotoWire testset are summarized in Table TABREF25. For each proposed variant of our architecture, we report the mean score over ten runs, as well as the standard deviation in subscript. Results are compared to baselines BIBREF12, BIBREF13, BIBREF10 and variants of our models. We also report the result of the oracle (metrics on the gold descriptions). Please note that gold descriptions trivially obtain 100% on all metrics expect RG, as they are all based on comparison with themselves. RG scores are different, as the IE system is imperfect and fails to extract accurate entities 4% of the time. RG-# is an absolute count.
Results ::: Ablation studies
To evaluate the impact of our model components, we first compare scenarios Flat, Hierarchical-k, and Hierarchical-kv. As shown in Table TABREF25, we can see the lower results obtained by the Flat scenario compared to the other scenarios (e.g. BLEU $16.7$ vs. $17.5$ for resp. Flat and Hierarchical-k), suggesting the effectiveness of encoding the data-structure using a hierarchy. This is expected, as losing explicit delimitation between entities makes it harder a) for the encoder to encode semantics of the objects contained in the table and b) for the attention mechanism to extract salient entities/records.
Second, the comparison between scenario Hierarchical-kv and Hierarchical-k shows that omitting entirely the influence of the record values in the attention mechanism is more effective: this last variant performs slightly better in all metrics excepted CS-R%, reinforcing our intuition that focusing on the structure modeling is an important part of data encoding as well as confirming the intuition explained in Section SECREF16: once an entity is selected, facts about this entity are relevant based on their key, not value which might add noise. To illustrate this intuition, we depict in Figure FIGREF27 attention scores (recall $\alpha _{i,t}$ and $\beta _{i,j,t}$ from equations (SECREF16) and (DISPLAY_FORM17)) for both variants Hierarchical-kv and Hierarchical-k. We particularly focus on the timestamp where the models should mention the number of points scored during the first quarter of the game. Scores of Hierarchical-k are sharp, with all of the weight on the correct record (PTS_QTR1, 26) whereas scores of Hierarchical-kv are more distributed over all PTS_QTR records, ultimately failing to retrieve the correct one.
Results ::: Comparison w.r.t. baselines.
From a general point of view, we can see from Table TABREF25 that our scenarios obtain significantly higher results in terms of BLEU over all models; our best model Hierarchical-k reaching $17.5$ vs. $16.5$ against the best baseline. This means that our models learns to generate fluent sequences of words, close to the gold descriptions, adequately picking up on domain lingo. Qualitative metrics are either better or on par with baselines. We show in Figure FIGREF29 a text generated by our best model, which can be directly compared to the gold description in Figure FIGREF1. Generation is fluent and contains domain-specific expressions. As reflected in Table TABREF25, the number of correct mentions (in green) outweights the number of incorrect mentions (in red). Please note that, as in previous work BIBREF28, BIBREF12, BIBREF13, BIBREF10, generated texts still contain a number of incorrect facts, as well hallucinations (in blue): sentences that have no basis in the input data (e.g. “[...] he's now averaging 22 points [...]."). While not the direct focus of our work, this highlights that any operation meant to enrich the semantics of structured data can also enrich the data with incorrect facts.
Specifically, regarding all baselines, we can outline the following statements.
$\bullet $ Our hierarchical models achieve significantly better scores on all metrics when compared to the flat architecture Wiseman, reinforcing the crucial role of structure in data semantics and saliency. The analysis of RG metrics shows that Wiseman seems to be the more naturalistic in terms of number of factual mentions (RG#) since it is the closest scenario to the gold value (16.83 vs. 17.31 for resp. Wiseman and Hierarchical-k). However, Wiseman achieves only $75.62$% of precision, effectively mentioning on average a total of $22.25$ records (wrong or accurate), where our model Hierarchical-k scores a precision of $89.46$%, leading to $23.66$ total mentions, just slightly above Wiseman.
$\bullet $ The comparison between the Flat scenario and Wiseman is particularly interesting. Indeed, these two models share the same intuition to flatten the data-structure. The only difference stands on the encoder mechanism: bi-LSTM vs. Transformer, for Wiseman and Flat respectively. Results shows that our Flat scenario obtains a significant higher BLEU score (16.7 vs. 14.5) and generates fluent descriptions with accurate mentions (RG-P%) that are also included in the gold descriptions (CS-R%). This suggests that introducing the Transformer architecture is promising way to implicitly account for data structure.
$\bullet $ Our hierarchical models outperform the two-step decoders of Li and Puduppully-plan on both BLEU and all qualitative metrics, showing that capturing structure in the encoding process is more effective that predicting a structure in the decoder (i.e., planning or templating). While our models sensibly outperform in precision at factual mentions, the baseline Puduppully-plan reaches $34.28$ mentions on average, showing that incorporating modules dedicated to entity extraction leads to over-focusing on entities; contrasting with our models that learn to generate more balanced descriptions.
$\bullet $ The comparison with Puduppully-updt shows that dynamically updating the encoding across the generation process can lead to better Content Ordering (CO) and RG-P%. However, this does not help with Content Selection (CS) since our best model Hierarchical-k obtains slightly better scores. Indeed, Puduppully-updt updates representations after each mention allowing to keep track of the mention history. This guides the ordering of mentions (CO metric), each step limiting more the number of candidate mentions (increasing RG-P%). In contrast, our model encodes saliency among records/entities more effectively (CS metric). We note that while our model encodes the data-structure once and for all, Puduppully-updt recomputes, via the updates, the encoding at each step and therefore significantly increases computation complexity. Combined with their RG-# score of $30.11$, we argue that our model is simpler, and obtains fluent description with accurate mentions in a more human-like fashion.
We would also like to draw attention to the number of parameters used by those architectures. We note that our scenarios relies on a lower number of parameters (14 millions) compared to all baselines (ranging from 23 to 45 millions). This outlines the effectiveness in the design of our model relying on a structure encoding, in contrast to other approach that try to learn the structure of data/descriptions from a linearized encoding.
Conclusion and future work
In this work we have proposed a hierarchical encoder for structured data, which 1) leverages the structure to form efficient representation of its input; 2) has strong synergy with the hierarchical attention of its associated decoder. This results in an effective and more light-weight model. Experimental evaluation on the RotoWire benchmark shows that our model outperforms competitive baselines in terms of BLEU score and is generally better on qualitative metrics. This way of representing structured databases may lead to automatic inference and enrichment, e.g., by comparing entities. This direction could be driven by very recent operation-guided networks BIBREF39, BIBREF40. In addition, we note that our approach can still lead to erroneous facts or even hallucinations. An interesting perspective might be to further constrain the model on the data structure in order to prevent inaccurate of even contradictory descriptions.
Acknowledgements
We would like to thank the H2020 project AI4EU (825619) which partially supports Laure Soulier and Patrick Gallinari.
Question:
Which qualitative metric are used for evaluation?
Answer:
| Relation Generation (RG), Content Selection (CS), Content Ordering (CO) | 328 | 31,546 | 31,548 | 31,607 | ... [The rest of the article is omitted]
|
76ed74788e3eb3321e646c48ae8bf6cdfe46dca1 | 76ed74788e3eb3321e646c48ae8bf6cdfe46dca1_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Modern Standard Arabic (MSA) and Classical Arabic (CA) have two types of vowels, namely long vowels, which are explicitly written, and short vowels, aka diacritics, which are typically omitted in writing but are reintroduced by readers to properly pronounce words. Since diacritics disambiguate the sense of the words in context and their syntactic roles in sentences, automatic diacritic recovery is essential for applications such as text-to-speech and educational tools for language learners, who may not know how to properly verbalize words. Diacritics have two types, namely: core-word (CW) diacritics, which are internal to words and specify lexical selection; and case-endings (CE), which appear on the last letter of word stems, typically specifying their syntactic role. For example, the word “ktb” (كتب>) can have multiple diacritized forms such as “katab” (كَتَب> – meaning “he wrote”) “kutub” (كُتُب> – “books”). While “katab” can only assume one CE, namely “fatHa” (“a”), “kutub” can accept the CEs: “damma” (“u”) (nominal – ex. subject), “a” (accusative – ex. object), “kasra” (“i”) (genitive – ex. PP predicate), or their nunations. There are 14 diacritic combinations. When used as CEs, they typically convey specific syntactic information, namely: fatHa “a” for accusative nouns, past verbs and subjunctive present verbs; kasra “i” for genitive nouns; damma “u” for nominative nouns and indicative present verbs; sukun “o” for jussive present verbs and imperative verbs. FatHa, kasra and damma can be preceded by shadda “$\sim $” for gemination (consonant doubling) and/or converted to nunation forms following some grammar rules. In addition, according to Arabic orthography and phonology, some words take a virtual (null) “#” marker when they end with certain characters (ex: long vowels). This applies also to all non-Arabic words (ex: punctuation, digits, Latin words, etc.). Generally, function words, adverbs and foreign named entities (NEs) have set CEs (sukun, fatHa or virtual). Similar to other Semitic languages, Arabic allows flexible Verb-Subject-Object as well as Verb-Object-Subject constructs BIBREF1. Such flexibility creates inherent ambiguity, which is resolved by diacritics as in “r$>$Y Emr Ely” (رأى عمر علي> Omar saw Ali/Ali saw Omar). In the absence of diacritics it is not clear who saw whom. Similarly, in the sub-sentence “kAn Alm&tmr AltAsE” (كان المؤتمر التاسع>), if the last word, is a predicate of the verb “kAn”, then the sentence would mean “this conference was the ninth” and would receive a fatHa (a) as a case ending. Conversely, if it was an adjective to the “conference”, then the sentence would mean “the ninth conference was ...” and would receive a damma (u) as a case ending. Thus, a consideration of context is required for proper disambiguation. Due to the inter-word dependence of CEs, they are typically harder to predict compared to core-word diacritics BIBREF2, BIBREF3, BIBREF4, BIBREF5, with CEER of state-of-the-art systems being in double digits compared to nearly 3% for word-cores. Since recovering CEs is akin to shallow parsing BIBREF6 and requires morphological and syntactic processing, it is a difficult problem in Arabic NLP. In this paper, we focus on recovering both CW diacritics and CEs. We employ two separate Deep Neural Network (DNN) architectures for recovering both kinds of diacritic types. We use character-level and word-level bidirectional Long-Short Term Memory (biLSTM) based recurrent neural models for CW diacritic and CE recovery respectively. We train models for both Modern Standard Arabic (MSA) and Classical Arabic (CA). For CW diacritics, the model is informed using word segmentation information and a unigram language model. We also employ a unigram language model to perform post correction on the model output. We achieve word error rates for CW diacritics of 2.9% and 2.2% for MSA and CA. The MSA word error rate is 6% lower than the best results in the literature (the RDI diacritizer BIBREF7). The CE model is trained with a rich set of surface, morphological, and syntactic features. The proposed features would aid the biLSTM model in capturing syntactic dependencies indicated by Part-Of-Speech (POS) tags, gender and number features, morphological patterns, and affixes. We show that our model achieves a case ending error rate (CEER) of 3.7% for MSA and 2.5% for CA. For MSA, this CEER is more than 60% lower than other state-of-the-art systems such as Farasa and the RDI diacritizer, which are trained on the same dataset and achieve CEERs of 10.7% and 14.4% respectively. The contributions of this paper are as follows:
We employ a character-level RNN model that is informed using word morphological information and a word unigram language model to recover CW diacritics. Our model beats the best state-of-the-art system by 6% for MSA.
We introduce a new feature rich RNN-based CE recovery model that achieves errors rates that are 60% lower than the current state-of-the-art for MSA.
We explore the effect of different features, which may potentially be exploited for Arabic parsing.
We show the effectiveness of our approach for both MSA and CA.
Background
Automatic diacritics restoration has been investigated for many different language such as European languages (e.g. Romanian BIBREF8, BIBREF9, French BIBREF10, and Croatian BIBREF11), African languages (e.g. Yorba BIBREF12), Southeast Asian languages (e.g. Vietnamese BIBREF13), Semitic language (e.g. Arabic and Hebrew BIBREF14), and many others BIBREF15. For many languages, diacritic (or accent restoration) is limited to a handful of letters. However, for Semitic languages, diacritic recovery extends to most letters. Many general approaches have been explored for this problem including linguistically motivated rule-based approaches, machine learning approaches, such as Hidden Markov Models (HMM) BIBREF14 and Conditional Random Fields (CRF) BIBREF16, and lately deep learning approaches such as Arabic BIBREF17, BIBREF18, BIBREF19, Slovak BIBREF20, and Yorba BIBREF12. Aside from rule-based approaches BIBREF21, different methods were used to recover diacritics in Arabic text. Using a hidden Markov model (HMM) BIBREF14, BIBREF22 with an input character sequence, the model attempts to find the best state sequence given previous observations. BIBREF14 reported a 14% word error rate (WER) while BIBREF22 achieved a 4.1% diacritic error rate (DER) on the Quran (CA). BIBREF23 combined both morphological, acoustic, and contextual features to build a diacritizer trained on FBIS and LDC CallHome ECA collections. They reported a 9% (DER) without CE, and 28% DER with CE. BIBREF24 employed a cascade of a finite state transducers. The cascade stacked a word language model (LM), a charachter LM, and a morphological model. The model achieved an accuracy of 7.33% WER without CE and and 23.61% WER with CE. BIBREF25 employed a maximum entropy model for sequence classification. The system was trained on the LDC’s Arabic Treebank (ATB) and evaluated on a 600 articles from An-Nahar Newspaper (340K words) and achieved 5.5% DER and 18% WER on words without CE.
BIBREF26 used a hybrid approach that utilizes the output of Alkhalil morphological Analyzer BIBREF27 to generate all possible out of context diacritizations of a word. Then, an HMM guesses the correct diacritized form. Similarly, Microsoft Arabic Toolkit Services (ATKS) diacritizer BIBREF28 uses a rule-based morphological analyzer that produces possible analyses and an HMM in conjunction with rules to guess the most likely analysis. They report WER of 11.4% and 4.4% with and without CE. MADAMIRA BIBREF29 uses a combinations of morpho-syntactic features to rank a list of potential analyses provided by the Buckwalter Arabic Morphological Analyzer (BAMA) BIBREF30. An SVM trained on ATB selects the most probable analysis, including the diacritized form. MADAMIRA achieves 19.0% and 6.7% WER with and without CE respectively BIBREF31. Farasa BIBREF31 uses an HMM to guess CW diacritics and an SVM-rank based model trained on morphological and syntactic features to guess CEs. Farasa achieves WER of 12.8% and 3.3% with and without CEs.
More recent work employed different neural architectures to model the diacritization problem. BIBREF17 used a biLSTM recurrent neural network trained on the same dataset as BIBREF25. They explored one, two and three BiLSTM layers with 250 nodes in each layers, achieving WER of 9.1% including CE on ATB. Similar architectures were used but achieved lower results BIBREF7, BIBREF32. BIBREF33 provide a comprehensive survey on Arabic diacritization. A more recent survey by BIBREF34 concluded that reported results are often incomparable due to the usage of different test sets. They concluded that a large unigram LM for CW diacritic recovery is competitive with many of the systems in the literature, which prompted us to utilize a unigram language model for post correction. As mentioned earlier, two conclusions can be drawn, namely: restoring CEs is more challenging than CW diacritic restoration; and combining multiple features typically improves CE restoration.
In this paper, we expand upon the work in the literature by introducing feature-rich DNN models for restoring both CW and CE diacritics. We compare our models to multiple systems on the same test set. We achieve results that reduce diacritization error rates by more than half compared to the best SOTA systems. We further conduct an ablation study to determine the relative effect of the different features. As for Arabic, it is a Semitic language with derivational morphology. Arabic nouns, adjectives, adverbs, and verbs are typically derived from a closed set of 10,000 roots of length 3, 4, or rarely 5. Arabic nouns and verbs are derived from roots by applying templates to the roots to generate stems. Such templates may carry information that indicate morphological features of words such POS tag, gender, and number. For example, given a 3-letter root with 3 consonants CCC, a valid template may be CwACC , where the infix “wA” (وا>) is inserted, this template typically indicates an Arabic broken, or irregular, plural template for a noun of template CACC or CACCp if masculine or feminine respectively. Further, stems may accept prefixes and/or suffixes to form words. Prefixes include coordinating conjunctions, determiner, and prepositions, and suffixes include attached pronouns and gender and number markers.
Our Diacritizer ::: Training and Test Corpora
For MSA, we acquired the diacritized corpus that was used to train the RDI BIBREF7 diacritizer and the Farasa diacritizer BIBREF31. The corpus contains 9.7M tokens with approximately 194K unique surface forms (excluding numbers and punctuation marks). The corpus covers multiple genres such as politics and sports and is a mix of MSA and CA. This corpus is considerably larger than the Arabic Treebank BIBREF35 and is more consistent in its diacritization. For testing, we used the freely available WikiNews test set BIBREF31, which is composed of 70 MSA WikiNews articles (18,300 tokens) and evenly covers a variety of genres including politics, economics, health, science and technology, sports, arts and culture.
For CA, we obtained a large collection of fully diacritized classical texts (2.7M tokens) from a book publisher, and we held-out a small subset of 5,000 sentences (approximately 400k words) for testing. Then, we used the remaining sentences to train the CA models.
Our Diacritizer ::: Core Word Diacritization ::: Features.
Arabic words are typically derived from a limited set of roots by fitting them into so-called stem-templates (producing stems) and may accept a variety of prefixes and suffixes such as prepositions, determiners, and pronouns (producing words). Word stems specify the lexical selection and are typically unaffected by the attached affixes. We used 4 feature types, namely:
CHAR: the characters.
SEG: the position of the character in a word segment. For example, given the word “wAlktAb” (والكتاب> and the book/writers), which is composed of 3 segments “w+Al+ktAb” (و+ال+كتاب>). Letters were marked as “B” if they begin a segment, “M” if they are in the middle of a segment, “E” if they end a segment, and “S” if they are single letter segments. So for “w+Al+ktAb”, the corresponding character positions are “S+BE+BMME”. We used Farasa to perform segmentation, which has a reported segmentation accuracy of 99% on the WikiNews dataset BIBREF36.
PRIOR: diacritics seen in the training set per segment. Since we used a character level model, this feature informed the model with word level information. For example, the word “ktAb” (كتاب>) was observed to have two diacritized forms in the training set, namely “kitaAb” (كِتَاب> – book) and “kut$\sim $aAb” (كُتَّاب> – writers). The first letter in the word (“k”) accepted the diacritics “i” and “u”. Thus given a binary vector representing whether a character is allowed to assume any of the eight primitive Arabic diacritic marks (a, i, u, o, K, N, F, and $\sim $ in order), the first letter would be given the following vector “01100000”. If a word segment was never observed during training, the vector for all letters therein would be set to 11111111. This feature borrows information from HMM models, which have been fairly successful in diacritizing word cores.
CASE: whether the letter expects a core word diacritic or a case ending. Case endings are placed on only one letter in a word, which may or may not be the last letter in the word. This is a binary feature.
Our Diacritizer ::: Core Word Diacritization ::: DNN Model.
Using a DNN model, particularly with a biLSTM BIBREF37, is advantageous because the model automatically explores the space of feature combinations and is able to capture distant dependencies. A number of studies have explored various biLSTM architectures BIBREF17, BIBREF7, BIBREF32 including stacked biLSTMs confirming their effectiveness. As shown in Figure FIGREF14, we employed a character-based biLSTM model with associated features for each character. Every input character had an associated list of $m$ features, and we trained randomly initialized embeddings of size 50 for each feature. Then, we concatenated the feature embeddings vectors creating an $m\times 50$ vector for each character, which was fed into the biLSTM layer of length 100. The output of the biLSTM layer was fed directly into a dense layer of size 100. We used early stopping with patience of 5 epochs, a learning rate of 0.001, a batch size of 256, and an Adamax optimizer. The input was the character sequence in a sentence with words being separated by word boundary markers (WB), and we set the maximum sentence length to 1,250 characters.
Our Diacritizer ::: Case Ending Diacritization ::: Features.
Table TABREF17 lists the features that we used for CE recovery. We used Farasa to perform segmentation and POS tagging and to determine stem-templates BIBREF31. Farasa has a reported POS accuracy of 96% on the WikiNews dataset BIBREF31. Though the Farasa diacritizer utilizes a combination of some the features presented herein, namely segmentation, POS tagging, and stem templates, Farasa's SVM-ranking approach requires explicit specification of feature combinations (ex. $Prob(CE\Vert current\_word, prev\_word, prev\_CE)$). Manual exploration of the feature space is undesirable, and ideally we would want our learning algorithm to do so automatically. The flexibility of the DNN model allowed us to include many more surface level features such as affixes, leading and trailing characters in words and stems, and the presence of words in large gazetteers of named entities. As we show later, these additional features significantly lowered CEER.
Our Diacritizer ::: Case Ending Diacritization ::: DNN Model
Figure FIGREF19 shows the architecture of our DNN algorithm. Every input word had an associated list of $n$ features, and we trained randomly initialized embeddings of size 100 for each feature. Then, we concatenated the feature embeddings vectors creating an $n\times 100$ vector for each word. We fed these vectors into a biLSTM layer of 100 dimensions after applying a dropout of 75%, where dropout behaves like a regularlizer to avoid overfitting BIBREF40. We conducted side experiments with lower dropout rates, but the higher dropout rate worked best. The output of the biLSTM layer was fed into a 100 dimensional dense layer with 15% dropout and softmax activation. We conducted side experiments where we added additional biLSTM layers and replaced softmax with a conditional random field layer, but we did not observe improvements. Thus, we opted for a simpler model. We used a validation set to determine optimal parameters such as dropout rate. Again, we used the “Adamax” optimizer with categorical cross entropy loss and a learning rate of 0.001. We also applied early stopping with patience of up to 5 consecutive epochs without improvement.
Experiments and Results ::: Core Word ::: Experimental Setup
For all the experiments conducted herein, we used the Keras toolkit BIBREF41 with a TensorFlow backend BIBREF42. We used the entirety of the training set as input, and we instructed Keras to use 5% of the data for tuning (validation). We included the CASE feature, which specifies whether the letter accepts a normal diacritic or case ending, in all our setups. We conducted multiple experiment using different features, namely:
CHAR: This is our baseline setup where we only used the characters as features.
CHAR+SEG: This takes the characters and their segmentation information as features.
CHAR+PRIOR: This takes the characters and their the observed diacritized forms in the training set.
All: This setup includes all the features.
We also optionally employed post correction. For words that were seen in training, if the model produced a diacritized form that was not seen in the training data, we assumed it was an error and replaced it with the most frequently observed diacritized form (using a unigram language model). We report two error rates, namely WER (at word level) and DER (at character level). We used relaxed scoring where we assumed an empty case to be equivalent to sukun, and we removed default diacritics – fatHa followed by alef, kasra followed by ya, and damma followed by wa. Using such scoring would allow to compare to other systems in the literature that may use different diacritization conventions.
Experiments and Results ::: Core Word ::: Results and Error analysis
For testing, we used the aforementioned WikiNews dataset to test the MSA diacritizer and the held-out 5,000 sentences for CA. Table TABREF27 shows WER and DER results using different features with and without post correction.
Experiments and Results ::: Core Word ::: Results and Error analysis ::: MSA Results:
For MSA, though the CHAR+PRIOR feature led to worse results than using CHAR alone, the results show that combining all the features achieved the best results. Moreover, post correction improved results overall. We compare our results to five other systems, namely Farasa BIBREF31, MADAMIRA BIBREF29, RDI (Rashwan et al., 2015), MIT (Belinkow and Glass, 2015), and Microsoft ATKS BIBREF28. Table TABREF34 compares our system with others in the aforementioned systems. As the results show, our results beat the current state-of-the-art.
For error analysis, we analyzed all the errors (527 errors). The errors types along with examples of each are shown in Table TABREF29. The most prominent error type arises from the selection of a valid diacritized form that does not match the context (40.8%). Perhaps, including POS tags as a feature or augmenting the PRIOR feature with POS tag information and a bigram language model may reduce the error rate further. The second most common error is due to transliterated foreign words including foreign named entities (23.5%). Such words were not observed during training. Further, Arabic Named entities account for 10.6% of the errors, where they were either not seen in training or they share identical non-diacritized forms with other words. Perhaps, building larger gazetteers of diacritized named entities may resolve NE related errors. In 10.8% of the cases, the diacritizer produced in completely incorrect diacritized forms. In some the cases (9.1%), though the diacritizer produced a form that is different from the reference, both forms were in fact correct. Most of these cases were due to variations in diacritization conventions (ex. “bare alef” (A) at start of a word receiving a diacritic or not). Other cases include foreign words and some words where both diacritized forms are equally valid.
Experiments and Results ::: Core Word ::: Results and Error analysis ::: CA Results:
For CA results, the CHAR+SEG and CHAR+PRIOR performed better than using characters alone with CHAR+PRIOR performing better than CHAR+SEG. As in the case with MSA, combining all the features led to the best results. Post correction had a significantly larger positive impact on results compared to what we observed for MSA. This may indicate that we need a larger training set. The best WER that we achieved for CW diacritics with post corrections is 2.2%. Since we did not have access to any publicly available system that is tuned for CA, we compared our best system to using our best MSA system to diacritize the CA test set, and the MSA diacritizer produced significantly lower results with a WER of 8.5% (see Table TABREF34). This highlights the large difference between MSA and CA and the need for systems that are specifically tuned for both.
We randomly selected and analyzed 500 errors (5.2% of the errors). The errors types along with examples of each are shown in Table TABREF33. The two most common errors involve the system producing completely correct diacritized forms (38.8%) or correct forms that don't match the context (31.4%). The relatively higher percentage of completely incorrect guesses, compared to MSA, may point to the higher lexical diversity of classical Arabic. As for MSA, we suspect that adding additional POS information and employing a word bigram to constrain the PRIOR feature may help reduce selection errors. Another prominent error is related to the diacritics that appear on attached suffixes, particularly pronouns, which depend on the choice of case ending (13.2%). Errors due to named entities are slightly fewer than those seen for MSA (8.8%). A noticeable number of mismatches between the guess and the reference are due to partial diacritization of the reference (4.4%). We plan to conduct an extra round of checks on the test set.
Experiments and Results ::: Case Ending ::: Experimental Setup
We conducted multiple experiments to determine the relative effect of the different features as follows:
word: This is our baseline setup, which uses word surface forms only.
word-surface: This setup uses the word surface forms, stems, prefixes, and suffixes (including noun suffixes). This simulates the case when no POS tagging information is available.
word-POS: This includes the word surface form and POS information, including gender and number of stems, prefixes, and suffixes.
word-morph: This includes words and their stem templates to capture morphological patterns.
word-surface-POS-morph: This setup uses all the features (surface, POS, and morphological).
all-misc: This uses all features plus word and stem leading and trailing character unigrams and bigrams in addition to sukun words and named entities.
For testing MSA, we used the aforementioned WikiNews dataset. Again, we compared our results to five other systems, namely Farasa BIBREF31, MADAMIRA BIBREF29, RDI (Rashwan et al., 2015), MIT (Belinkow and Glass, 2015), and Microsoft ATKS BIBREF28. For CA testing, we used the 5,000 sentences that we set aside. Again, we compared to our best MSA system.
Experiments and Results ::: Case Ending ::: Results and Error Analysis
Table TABREF45 lists the results of our setups compared to other systems.
Experiments and Results ::: Case Ending ::: Results and Error Analysis ::: MSA Results:
As the results show, our baseline DNN system outperforms all state-of-the-art systems. Further, adding more features yielded better results overall. Surface-level features resulted in the most gain, followed by POS tags, and lastly stem templates. Further, adding head and tail characters along with a list of sukun words and named entities led to further improvement. Our proposed feature-rich system has a CEER that is approximately 61% lower than any of the state-of-the-art systems.
Figure FIGREF49 shows CE distribution and prediction accuracy. For the four basic markers kasra, fatHa, damma and sukun, which appear 27%, 14%, 9% and 10% respectively, the system has CEER of $\sim $1% for each. Detecting the virtual CE mark is a fairly easy task. All other CE markers represent 13% with almost negligible errors. Table TABREF30 lists a thorough breakdown of all errors accounting for at 1% of the errors along with the most common reasons of the errors and examples illustrating these reasons. For example, the most common error type involves guessing a fatHa (a) instead of damma (u) or vice versa (19.3%). The most common reasons for this error type, based on inspecting the errors, were due to: POS errors (ex. a word is tagged as a verb instead of a noun); and a noun is treated as a subject instead of an object or vice versa. The table details the rest of the error types. Overall, some of the errors are potentially fixable using better POS tagging, improved detection of non-Arabized foreign names, and detection of indeclinability. However, some errors are more difficult and require greater understanding of semantics such as improper attachment, incorrect idafa, and confusion between subject and object. Perhaps, such semantic errors can be resolved using parsing.
Experiments and Results ::: Case Ending ::: Results and Error Analysis ::: CA Results:
The results show that the POS tagging features led to the most improvements followed by the surface features. Combining all features led to the best results with WER of 2.5%. As we saw for CW diacritics, using our best MSA system to diacritize CA led to significantly lower results with CEER of 8.9%.
Figure FIGREF50 shows CE distribution and prediction accuracy. For the four basic markers fatHa, kasra, damma and sukun, which appear 18%, 14%, 13% and 8% respectively, the system has CEER $\sim $0.5% for each. Again, detecting the virtual CE mark was a fairly easy task. All other CE markers representing 20% have negligible errors.
Table TABREF31 lists all the error types, which account for at least 1% of the errors, along with their most common causes and explanatory examples. The error types are similar to those observed for MSA. Some errors are more syntactic and morphological in nature and can be addressed using better POS tagging and identification of indeclinability, particularly as they relate to named entities and nouns with feminine markers. Other errors such as incorrect attachment, incorrect idafa, false subject, and confusion between subject and object can perhaps benefit from the use of parsing. As with the core-word errors for CA, the reference has some errors (ex. {a,i,o} $\Rightarrow $ #), and extra rounds of reviews of the reference are in order.
Experiments and Results ::: Full Diacritization Results
Table TABREF48 compares the full word diacritization (CW+CE) of our best setup to other systems in the literature. As the results show for MSA, our overall diacritization WER is 6.0% while the state of the art system has a WER of 12.2%. As for CA, our best system produced an error rate of 4.3%, which is significantly better than using our best MSA system to diacritize CA.
Conclusion and Future Work
In this paper, we presented a feature-rich DNN approach for MSA CW and CE recovery that produces a word level error for MSA of 6.0%, which is more than 50% lower than state-of-the-art systems (6.0% compared to 12.2%) and word error rate of 4.3% for CA. Specifically, we used biLSTM-based model with a variety of surface, morphological, and syntactic features. Reliable NLP tools may be required to generate some of these features, and such tools may not be readily available for other language varieties, such as dialectal Arabic. However, we showed the efficacy of different varieties of features, such as surface level-features, and they can help improve diacritization individually. Further, though some errors may be overcome using improved NLP tools (ex. better POS tagging), semantic errors, such incorrect attchment, are more difficult to fix. Perhaps, using dependency parsing may help overcome some semantic errors. As for feature engineering, the broad categories of features, such as surface, syntactic, and morphological features, may likely carry-over to other languages, language-specific feature engineering may be require to handle the specificity of each language. Lastly, since multiple diacritization conventions may exist, as in the case of Arabic, adopting one convention consistently is important for training a good system and for properly testing it. Though we have mostly achieved this for MSA, the CA dataset requires more checks to insure greater consistency.
For future work, we want to explore the effectiveness of augmenting our CW model with POS tagging information and a bigram language model. Further, we plan to create a multi reference diacritization test set to handle words that have multiple valid diacritized forms. For CE, we want to examine the effectiveness of the proposed features for Arabic parsing. We plan to explore: character-level convolutional neural networks that may capture sub-word morphological features; pre-trained embeddings; and attention mechanisms to focus on salient features. We also plan to explore joint modeling for both core word and case ending diacritics.
Question:
what linguistics features are used?
Answer:
| POS, gender/number and stem POS | 328 | 30,319 | 30,321 | 30,366 | ... [The rest of the article is omitted]
|
8e52637026bee9061f9558178eaec08279bf7ac6 | 8e52637026bee9061f9558178eaec08279bf7ac6_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Understanding the emotions expressed in a text or message is of high relevance nowadays. Companies are interested in this to get an understanding of the sentiment of their current customers regarding their products and the sentiment of their potential customers to attract new ones. Moreover, changes in a product or a company may also affect the sentiment of a customer. However, the intensity of an emotion is crucial in determining the urgency and importance of that sentiment. If someone is only slightly happy about a product, is a customer willing to buy it again? Conversely, if someone is very angry about customer service, his or her complaint might be given priority over somewhat milder complaints.
BIBREF0 present four tasks in which systems have to automatically determine the intensity of emotions (EI) or the intensity of the sentiment (Valence) of tweets in the languages English, Arabic, and Spanish. The goal is to either predict a continuous regression (reg) value or to do ordinal classification (oc) based on a number of predefined categories. The EI tasks have separate training sets for four different emotions: anger, fear, joy and sadness. Due to the large number of subtasks and the fact that this language does not have many resources readily available, we only focus on the Spanish subtasks. Our work makes the following contributions:
Our submissions ranked second (EI-Reg), second (EI-Oc), fourth (V-Reg) and fifth (V-Oc), demonstrating that the proposed method is accurate in automatically determining the intensity of emotions and sentiment of Spanish tweets. This paper will first focus on the datasets, the data generation procedure, and the techniques and tools used. Then we present the results in detail, after which we perform a small error analysis on the largest mistakes our model made. We conclude with some possible ideas for future work.
Data
For each task, the training data that was made available by the organizers is used, which is a selection of tweets with for each tweet a label describing the intensity of the emotion or sentiment BIBREF1 . Links and usernames were replaced by the general tokens URL and @username, after which the tweets were tokenized by using TweetTokenizer. All text was lowercased. In a post-processing step, it was ensured that each emoji is tokenized as a single token.
Word Embeddings
To be able to train word embeddings, Spanish tweets were scraped between November 8, 2017 and January 12, 2018. We chose to create our own embeddings instead of using pre-trained embeddings, because this way the embeddings would resemble the provided data set: both are based on Twitter data. Added to this set was the Affect in Tweets Distant Supervision Corpus (DISC) made available by the organizers BIBREF0 and a set of 4.1 million tweets from 2015, obtained from BIBREF2 . After removing duplicate tweets and tweets with fewer than ten tokens, this resulted in a set of 58.7 million tweets, containing 1.1 billion tokens. The tweets were preprocessed using the method described in Section SECREF6 . The word embeddings were created using word2vec in the gensim library BIBREF3 , using CBOW, a window size of 40 and a minimum count of 5. The feature vectors for each tweet were then created by using the AffectiveTweets WEKA package BIBREF4 .
Translating Lexicons
Most lexical resources for sentiment analysis are in English. To still be able to benefit from these sources, the lexicons in the AffectiveTweets package were translated to Spanish, using the machine translation platform Apertium BIBREF5 .
All lexicons from the AffectiveTweets package were translated, except for SentiStrength. Instead of translating this lexicon, the English version was replaced by the Spanish variant made available by BIBREF6 .
For each subtask, the optimal combination of lexicons was determined. This was done by first calculating the benefits of adding each lexicon individually, after which only beneficial lexicons were added until the score did not increase anymore (e.g. after adding the best four lexicons the fifth one did not help anymore, so only four were added). The tests were performed using a default SVM model, with the set of word embeddings described in the previous section. Each subtask thus uses a different set of lexicons (see Table TABREF1 for an overview of the lexicons used in our final ensemble). For each subtask, this resulted in a (modest) increase on the development set, between 0.01 and 0.05.
Translating Data
The training set provided by BIBREF0 is not very large, so it was interesting to find a way to augment the training set. A possible method is to simply translate the datasets into other languages, leaving the labels intact. Since the present study focuses on Spanish tweets, all tweets from the English datasets were translated into Spanish. This new set of “Spanish” data was then added to our original training set. Again, the machine translation platform Apertium BIBREF5 was used for the translation of the datasets.
Algorithms Used
Three types of models were used in our system, a feed-forward neural network, an LSTM network and an SVM regressor. The neural nets were inspired by the work of Prayas BIBREF7 in the previous shared task. Different regression algorithms (e.g. AdaBoost, XGBoost) were also tried due to the success of SeerNet BIBREF8 , but our study was not able to reproduce their results for Spanish.
For both the LSTM network and the feed-forward network, a parameter search was done for the number of layers, the number of nodes and dropout used. This was done for each subtask, i.e. different tasks can have a different number of layers. All models were implemented using Keras BIBREF9 . After the best parameter settings were found, the results of 10 system runs to produce our predictions were averaged (note that this is different from averaging our different type of models in Section SECREF16 ). For the SVM (implemented in scikit-learn BIBREF10 ), the RBF kernel was used and a parameter search was conducted for epsilon. Detailed parameter settings for each subtask are shown in Table TABREF12 . Each parameter search was performed using 10-fold cross validation, as to not overfit on the development set.
Semi-supervised Learning
One of the aims of this study was to see if using semi-supervised learning is beneficial for emotion intensity tasks. For this purpose, the DISC BIBREF0 corpus was used. This corpus was created by querying certain emotion-related words, which makes it very suitable as a semi-supervised corpus. However, the specific emotion the tweet belonged to was not made public. Therefore, a method was applied to automatically assign the tweets to an emotion by comparing our scraped tweets to this new data set.
First, in an attempt to obtain the query-terms, we selected the 100 words which occurred most frequently in the DISC corpus, in comparison with their frequencies in our own scraped tweets corpus. Words that were clearly not indicators of emotion were removed. The rest was annotated per emotion or removed if it was unclear to which emotion the word belonged. This allowed us to create silver datasets per emotion, assigning tweets to an emotion if an annotated emotion-word occurred in the tweet.
Our semi-supervised approach is quite straightforward: first a model is trained on the training set and then this model is used to predict the labels of the silver data. This silver data is then simply added to our training set, after which the model is retrained. However, an extra step is applied to ensure that the silver data is of reasonable quality. Instead of training a single model initially, ten different models were trained which predict the labels of the silver instances. If the highest and lowest prediction do not differ more than a certain threshold the silver instance is maintained, otherwise it is discarded.
This results in two parameters that could be optimized: the threshold and the number of silver instances that would be added. This method can be applied to both the LSTM and feed-forward networks that were used. An overview of the characteristics of our data set with the final parameter settings is shown in Table TABREF14 . Usually, only a small subset of data was added to our training set, meaning that most of the silver data is not used in the experiments. Note that since only the emotions were annotated, this method is only applicable to the EI tasks.
Ensembling
To boost performance, the SVM, LSTM, and feed-forward models were combined into an ensemble. For both the LSTM and feed-forward approach, three different models were trained. The first model was trained on the training data (regular), the second model was trained on both the training and translated training data (translated) and the third one was trained on both the training data and the semi-supervised data (silver). Due to the nature of the SVM algorithm, semi-supervised learning does not help, so only the regular and translated model were trained in this case. This results in 8 different models per subtask. Note that for the valence tasks no silver training data was obtained, meaning that for those tasks the semi-supervised models could not be used.
Per task, the LSTM and feed-forward model's predictions were averaged over 10 prediction runs. Subsequently, the predictions of all individual models were combined into an average. Finally, models were removed from the ensemble in a stepwise manner if the removal increased the average score. This was done based on their original scores, i.e. starting out by trying to remove the worst individual model and working our way up to the best model. We only consider it an increase in score if the difference is larger than 0.002 (i.e. the difference between 0.716 and 0.718). If at some point the score does not increase and we are therefore unable to remove a model, the process is stopped and our best ensemble of models has been found. This process uses the scores on the development set of different combinations of models. Note that this means that the ensembles for different subtasks can contain different sets of models. The final model selections can be found in Table TABREF17 .
Results and Discussion
Table TABREF18 shows the results on the development set of all individuals models, distinguishing the three types of training: regular (r), translated (t) and semi-supervised (s). In Tables TABREF17 and TABREF18 , the letter behind each model (e.g. SVM-r, LSTM-r) corresponds to the type of training used. Comparing the regular and translated columns for the three algorithms, it shows that in 22 out of 30 cases, using translated instances as extra training data resulted in an improvement. For the semi-supervised learning approach, an improvement is found in 15 out of 16 cases. Moreover, our best individual model for each subtask (bolded scores in Table TABREF18 ) is always either a translated or semi-supervised model. Table TABREF18 also shows that, in general, our feed-forward network obtained the best results, having the highest F-score for 8 out of 10 subtasks.
However, Table TABREF19 shows that these scores can still be improved by averaging or ensembling the individual models. On the dev set, averaging our 8 individual models results in a better score for 8 out of 10 subtasks, while creating an ensemble beats all of the individual models as well as the average for each subtask. On the test set, however, only a small increase in score (if any) is found for stepwise ensembling, compared to averaging. Even though the results do not get worse, we cannot conclude that stepwise ensembling is a better method than simply averaging.
Our official scores (column Ens Test in Table TABREF19 ) have placed us second (EI-Reg, EI-Oc), fourth (V-Reg) and fifth (V-Oc) on the SemEval AIT-2018 leaderboard. However, it is evident that the results obtained on the test set are not always in line with those achieved on the development set. Especially on the anger subtask for both EI-Reg and EI-Oc, the scores are considerably lower on the test set in comparison with the results on the development set. Therefore, a small error analysis was performed on the instances where our final model made the largest errors.
Error Analysis
Due to some large differences between our results on the dev and test set of this task, we performed a small error analysis in order to see what caused these differences. For EI-Reg-anger, the gold labels were compared to our own predictions, and we manually checked 50 instances for which our system made the largest errors.
Some examples that were indicative of the shortcomings of our system are shown in Table TABREF20 . First of all, our system did not take into account capitalization. The implications of this are shown in the first sentence, where capitalization intensifies the emotion used in the sentence. In the second sentence, the name Imperator Furiosa is not understood. Since our texts were lowercased, our system was unable to capture the named entity and thought the sentence was about an angry emperor instead. In the third sentence, our system fails to capture that when you are so angry that it makes you laugh, it results in a reduced intensity of the angriness. Finally, in the fourth sentence, it is the figurative language me infla la vena (it inflates my vein) that the system is not able to understand.
The first two error-categories might be solved by including smart features regarding capitalization and named entity recognition. However, the last two categories are problems of natural language understanding and will be very difficult to fix.
Conclusion
To conclude, the present study described our submission for the Semeval 2018 Shared Task on Affect in Tweets. We participated in four Spanish subtasks and our submissions ranked second, second, fourth and fifth place. Our study aimed to investigate whether the automatic generation of additional training data through translation and semi-supervised learning, as well as the creation of stepwise ensembles, increase the performance of our Spanish-language models. Strong support was found for the translation and semi-supervised learning approaches; our best models for all subtasks use either one of these approaches. These results suggest that both of these additional data resources are beneficial when determining emotion intensity (for Spanish). However, the creation of a stepwise ensemble from the best models did not result in better performance compared to simply averaging the models. In addition, some signs of overfitting on the dev set were found. In future work, we would like to apply the methods (translation and semi-supervised learning) used on Spanish on other low-resource languages and potentially also on other tasks.
Question:
How was the training data translated?
Answer:
| using the machine translation platform Apertium | 328 | 15,139 | 15,141 | 15,188 | ... [The rest of the article is omitted]
|
8e52637026bee9061f9558178eaec08279bf7ac6 | 8e52637026bee9061f9558178eaec08279bf7ac6_1 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Understanding the emotions expressed in a text or message is of high relevance nowadays. Companies are interested in this to get an understanding of the sentiment of their current customers regarding their products and the sentiment of their potential customers to attract new ones. Moreover, changes in a product or a company may also affect the sentiment of a customer. However, the intensity of an emotion is crucial in determining the urgency and importance of that sentiment. If someone is only slightly happy about a product, is a customer willing to buy it again? Conversely, if someone is very angry about customer service, his or her complaint might be given priority over somewhat milder complaints.
BIBREF0 present four tasks in which systems have to automatically determine the intensity of emotions (EI) or the intensity of the sentiment (Valence) of tweets in the languages English, Arabic, and Spanish. The goal is to either predict a continuous regression (reg) value or to do ordinal classification (oc) based on a number of predefined categories. The EI tasks have separate training sets for four different emotions: anger, fear, joy and sadness. Due to the large number of subtasks and the fact that this language does not have many resources readily available, we only focus on the Spanish subtasks. Our work makes the following contributions:
Our submissions ranked second (EI-Reg), second (EI-Oc), fourth (V-Reg) and fifth (V-Oc), demonstrating that the proposed method is accurate in automatically determining the intensity of emotions and sentiment of Spanish tweets. This paper will first focus on the datasets, the data generation procedure, and the techniques and tools used. Then we present the results in detail, after which we perform a small error analysis on the largest mistakes our model made. We conclude with some possible ideas for future work.
Data
For each task, the training data that was made available by the organizers is used, which is a selection of tweets with for each tweet a label describing the intensity of the emotion or sentiment BIBREF1 . Links and usernames were replaced by the general tokens URL and @username, after which the tweets were tokenized by using TweetTokenizer. All text was lowercased. In a post-processing step, it was ensured that each emoji is tokenized as a single token.
Word Embeddings
To be able to train word embeddings, Spanish tweets were scraped between November 8, 2017 and January 12, 2018. We chose to create our own embeddings instead of using pre-trained embeddings, because this way the embeddings would resemble the provided data set: both are based on Twitter data. Added to this set was the Affect in Tweets Distant Supervision Corpus (DISC) made available by the organizers BIBREF0 and a set of 4.1 million tweets from 2015, obtained from BIBREF2 . After removing duplicate tweets and tweets with fewer than ten tokens, this resulted in a set of 58.7 million tweets, containing 1.1 billion tokens. The tweets were preprocessed using the method described in Section SECREF6 . The word embeddings were created using word2vec in the gensim library BIBREF3 , using CBOW, a window size of 40 and a minimum count of 5. The feature vectors for each tweet were then created by using the AffectiveTweets WEKA package BIBREF4 .
Translating Lexicons
Most lexical resources for sentiment analysis are in English. To still be able to benefit from these sources, the lexicons in the AffectiveTweets package were translated to Spanish, using the machine translation platform Apertium BIBREF5 .
All lexicons from the AffectiveTweets package were translated, except for SentiStrength. Instead of translating this lexicon, the English version was replaced by the Spanish variant made available by BIBREF6 .
For each subtask, the optimal combination of lexicons was determined. This was done by first calculating the benefits of adding each lexicon individually, after which only beneficial lexicons were added until the score did not increase anymore (e.g. after adding the best four lexicons the fifth one did not help anymore, so only four were added). The tests were performed using a default SVM model, with the set of word embeddings described in the previous section. Each subtask thus uses a different set of lexicons (see Table TABREF1 for an overview of the lexicons used in our final ensemble). For each subtask, this resulted in a (modest) increase on the development set, between 0.01 and 0.05.
Translating Data
The training set provided by BIBREF0 is not very large, so it was interesting to find a way to augment the training set. A possible method is to simply translate the datasets into other languages, leaving the labels intact. Since the present study focuses on Spanish tweets, all tweets from the English datasets were translated into Spanish. This new set of “Spanish” data was then added to our original training set. Again, the machine translation platform Apertium BIBREF5 was used for the translation of the datasets.
Algorithms Used
Three types of models were used in our system, a feed-forward neural network, an LSTM network and an SVM regressor. The neural nets were inspired by the work of Prayas BIBREF7 in the previous shared task. Different regression algorithms (e.g. AdaBoost, XGBoost) were also tried due to the success of SeerNet BIBREF8 , but our study was not able to reproduce their results for Spanish.
For both the LSTM network and the feed-forward network, a parameter search was done for the number of layers, the number of nodes and dropout used. This was done for each subtask, i.e. different tasks can have a different number of layers. All models were implemented using Keras BIBREF9 . After the best parameter settings were found, the results of 10 system runs to produce our predictions were averaged (note that this is different from averaging our different type of models in Section SECREF16 ). For the SVM (implemented in scikit-learn BIBREF10 ), the RBF kernel was used and a parameter search was conducted for epsilon. Detailed parameter settings for each subtask are shown in Table TABREF12 . Each parameter search was performed using 10-fold cross validation, as to not overfit on the development set.
Semi-supervised Learning
One of the aims of this study was to see if using semi-supervised learning is beneficial for emotion intensity tasks. For this purpose, the DISC BIBREF0 corpus was used. This corpus was created by querying certain emotion-related words, which makes it very suitable as a semi-supervised corpus. However, the specific emotion the tweet belonged to was not made public. Therefore, a method was applied to automatically assign the tweets to an emotion by comparing our scraped tweets to this new data set.
First, in an attempt to obtain the query-terms, we selected the 100 words which occurred most frequently in the DISC corpus, in comparison with their frequencies in our own scraped tweets corpus. Words that were clearly not indicators of emotion were removed. The rest was annotated per emotion or removed if it was unclear to which emotion the word belonged. This allowed us to create silver datasets per emotion, assigning tweets to an emotion if an annotated emotion-word occurred in the tweet.
Our semi-supervised approach is quite straightforward: first a model is trained on the training set and then this model is used to predict the labels of the silver data. This silver data is then simply added to our training set, after which the model is retrained. However, an extra step is applied to ensure that the silver data is of reasonable quality. Instead of training a single model initially, ten different models were trained which predict the labels of the silver instances. If the highest and lowest prediction do not differ more than a certain threshold the silver instance is maintained, otherwise it is discarded.
This results in two parameters that could be optimized: the threshold and the number of silver instances that would be added. This method can be applied to both the LSTM and feed-forward networks that were used. An overview of the characteristics of our data set with the final parameter settings is shown in Table TABREF14 . Usually, only a small subset of data was added to our training set, meaning that most of the silver data is not used in the experiments. Note that since only the emotions were annotated, this method is only applicable to the EI tasks.
Ensembling
To boost performance, the SVM, LSTM, and feed-forward models were combined into an ensemble. For both the LSTM and feed-forward approach, three different models were trained. The first model was trained on the training data (regular), the second model was trained on both the training and translated training data (translated) and the third one was trained on both the training data and the semi-supervised data (silver). Due to the nature of the SVM algorithm, semi-supervised learning does not help, so only the regular and translated model were trained in this case. This results in 8 different models per subtask. Note that for the valence tasks no silver training data was obtained, meaning that for those tasks the semi-supervised models could not be used.
Per task, the LSTM and feed-forward model's predictions were averaged over 10 prediction runs. Subsequently, the predictions of all individual models were combined into an average. Finally, models were removed from the ensemble in a stepwise manner if the removal increased the average score. This was done based on their original scores, i.e. starting out by trying to remove the worst individual model and working our way up to the best model. We only consider it an increase in score if the difference is larger than 0.002 (i.e. the difference between 0.716 and 0.718). If at some point the score does not increase and we are therefore unable to remove a model, the process is stopped and our best ensemble of models has been found. This process uses the scores on the development set of different combinations of models. Note that this means that the ensembles for different subtasks can contain different sets of models. The final model selections can be found in Table TABREF17 .
Results and Discussion
Table TABREF18 shows the results on the development set of all individuals models, distinguishing the three types of training: regular (r), translated (t) and semi-supervised (s). In Tables TABREF17 and TABREF18 , the letter behind each model (e.g. SVM-r, LSTM-r) corresponds to the type of training used. Comparing the regular and translated columns for the three algorithms, it shows that in 22 out of 30 cases, using translated instances as extra training data resulted in an improvement. For the semi-supervised learning approach, an improvement is found in 15 out of 16 cases. Moreover, our best individual model for each subtask (bolded scores in Table TABREF18 ) is always either a translated or semi-supervised model. Table TABREF18 also shows that, in general, our feed-forward network obtained the best results, having the highest F-score for 8 out of 10 subtasks.
However, Table TABREF19 shows that these scores can still be improved by averaging or ensembling the individual models. On the dev set, averaging our 8 individual models results in a better score for 8 out of 10 subtasks, while creating an ensemble beats all of the individual models as well as the average for each subtask. On the test set, however, only a small increase in score (if any) is found for stepwise ensembling, compared to averaging. Even though the results do not get worse, we cannot conclude that stepwise ensembling is a better method than simply averaging.
Our official scores (column Ens Test in Table TABREF19 ) have placed us second (EI-Reg, EI-Oc), fourth (V-Reg) and fifth (V-Oc) on the SemEval AIT-2018 leaderboard. However, it is evident that the results obtained on the test set are not always in line with those achieved on the development set. Especially on the anger subtask for both EI-Reg and EI-Oc, the scores are considerably lower on the test set in comparison with the results on the development set. Therefore, a small error analysis was performed on the instances where our final model made the largest errors.
Error Analysis
Due to some large differences between our results on the dev and test set of this task, we performed a small error analysis in order to see what caused these differences. For EI-Reg-anger, the gold labels were compared to our own predictions, and we manually checked 50 instances for which our system made the largest errors.
Some examples that were indicative of the shortcomings of our system are shown in Table TABREF20 . First of all, our system did not take into account capitalization. The implications of this are shown in the first sentence, where capitalization intensifies the emotion used in the sentence. In the second sentence, the name Imperator Furiosa is not understood. Since our texts were lowercased, our system was unable to capture the named entity and thought the sentence was about an angry emperor instead. In the third sentence, our system fails to capture that when you are so angry that it makes you laugh, it results in a reduced intensity of the angriness. Finally, in the fourth sentence, it is the figurative language me infla la vena (it inflates my vein) that the system is not able to understand.
The first two error-categories might be solved by including smart features regarding capitalization and named entity recognition. However, the last two categories are problems of natural language understanding and will be very difficult to fix.
Conclusion
To conclude, the present study described our submission for the Semeval 2018 Shared Task on Affect in Tweets. We participated in four Spanish subtasks and our submissions ranked second, second, fourth and fifth place. Our study aimed to investigate whether the automatic generation of additional training data through translation and semi-supervised learning, as well as the creation of stepwise ensembles, increase the performance of our Spanish-language models. Strong support was found for the translation and semi-supervised learning approaches; our best models for all subtasks use either one of these approaches. These results suggest that both of these additional data resources are beneficial when determining emotion intensity (for Spanish). However, the creation of a stepwise ensemble from the best models did not result in better performance compared to simply averaging the models. In addition, some signs of overfitting on the dev set were found. In future work, we would like to apply the methods (translation and semi-supervised learning) used on Spanish on other low-resource languages and potentially also on other tasks.
Question:
How was the training data translated?
Answer:
| machine translation platform Apertium BIBREF5 | 328 | 15,139 | 15,141 | 15,188 | ... [The rest of the article is omitted]
|
3116453e35352a3a90ee5b12246dc7f2e60cfc59 | 3116453e35352a3a90ee5b12246dc7f2e60cfc59_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Medical text mining is an exciting area and is becoming attractive to natural language processing (NLP) researchers. Clinical notes are an example of text in the medical area that recent work has focused on BIBREF0, BIBREF1, BIBREF2. This work studies abbreviation disambiguation on clinical notes BIBREF3, BIBREF4, specifically those used commonly by physicians and nurses. Such clinical abbreviations can have a large number of meanings, depending on the specialty BIBREF5, BIBREF6. For example, the term MR can mean magnetic resonance, mitral regurgitation, mental retardation, medical record and the general English Mister (Mr.). Table TABREF1 illustrates such an example. Abbreviation disambiguation is an important task in medical text understanding task BIBREF7. Successful recognition of the abbreviations in the notes can contribute to downstream tasks such as classification, named entity recognition, and relation extraction BIBREF7.
Recent work focuses on formulating the abbreviation disambiguation task as a classification problem, where the possible senses of a given abbreviation term are pre-defined with the help of domain experts BIBREF6, BIBREF5. Traditional features such as part-of-speech (POS) and Term Frequency-Inverse Document Frequency (TF-IDF) are widely investigated for clinical notes classification. Classifiers like support vector machines (SVMs) and random forests (RFs) are used to make predictions BIBREF1. Such methods depend heavily on feature engineering. Recently, deep features have been investigated in the medical domain. Word embeddings BIBREF8 and Convolutional Neural Networks (CNNs) BIBREF9, BIBREF10 provide very competitive performance on text classification for clinical notes and abbreviation disambiguation task BIBREF0, BIBREF11, BIBREF12, BIBREF6. Beyond vanilla embeddings, BIBREF13 utilized contextual features to do abbreviation expansion. Another challenge is the difficulty in obtaining training data: clinical notes have many restrictions due to privacy issues and require domain experts to develop high-quality annotations, thus leading to limited annotated training and testing data. Another difficulty is that in the real world (and in the existing public datasets), some abbreviation term-sense pairs (for example, AB as abortion) have a very high frequency of occurrence BIBREF5, while others are rarely found. This long tail issue creates the challenge of training under unbalanced sample distributions. We tackle these problems in the setting of few-shot learning BIBREF14, BIBREF15 where only a few or a low number of samples can be found in the training dataset to make use of limited resources. We propose a model that combines deep contextual features based on ELMo BIBREF16 and topic information to solve the abbreviation disambiguation task.
Our contributions can be summarized as: 1) we re-examined and corrected an existing dataset for training and we collected a test set for evaluation with focus especially for rare senses; 2) we proposed a few-shot learning approach which combines topic information and contextualized word embeddings to solve clinical abbreviation disambiguation task. The implementation is available online; 3) as limited research are conducted on this particular task, we evaluated and compared a number of baseline methods including classical models and deep models comprehensively.
Datasets
Training Dataset UM Inventory BIBREF5 is a public dataset created by researchers from the University of Minnesota, containing about 37,500 training samples with 75 abbreviation terms. Existing work reports abbreviation disambiguation results on 50 abbreviation terms BIBREF6, BIBREF5, BIBREF17. However, after carefully reviewing this dataset, we found that it contains many samples where medical professionals disagree: wrong samples and uncategorized samples. Due to these mistakes and flaws of this dataset, we removed the erroneous samples and eventually selected 30 abbreviation terms as our training dataset that can be made public. Among all the abbreviation terms, we have 11,466 samples, and 93 term-sense pairs in total (on average 123.3 samples/term-sense pair and 382.2 samples/term). Some term-sense pairs are very popular with a larger number of training samples but some are not (typically less than 5), we call them rare-sense cases . More details can be found in Appendix SECREF7.
Testing Dataset Our testing dataset takes MIMIC-III BIBREF18 and PubMed as data sources. Here we are referring to all the notes data from MIMIC-III (NOTEEVENTS table ) and Case Reports articles from PubMed, as this type of contents are close to medical notes. We provide detailed information in Appendix SECREF8, including the steps to create the testing dataset. Eventually, we have a balanced testing dataset, where each term-sense pair has at least 11 and up to 15 samples for training (on average, each pair has 14.56 samples and the median sample number is 15).
Baselines
We conducted a comprehensive comparison with the baseline models, and some of them were never investigated for the abbreviation disambiguation task. We applied traditional features by simply taking the TF-IDF features as the inputs into the classic classifiers. Deep features are also considered: a Doc2vec model BIBREF19 was pre-trained using Gensim and these word embeddings were applied to initialize deep models and fine-tuned.
TF-IDF: We applied TF-IDF features as inputs to four classifiers: support vector machine classifier (SVM), logistic regression classifier (LR), Naive Bayes classifier (NB) and random forest (RF); CNN: We followed the same architecture as BIBREF9; LSTM: We applied an LSTM model BIBREF20 to classify the sentences with pre-trained word embeddings;LSTM-soft: We then added a soft-attention BIBREF21 layer on top of the LSTM model where we computed soft alignment scores over each of the hidden states; LSTM-self: We applied a self-attention layer BIBREF22 to LSTM model. We denote the content vector as $c_i$ for each sentence $i$, as the input to the last classification layer.
Topic-attention Model
ELMo ELMo is a new type of word representation introduced by BIBREF16 which considers contextual information captured by pre-trained BiLSTM language models. Similar works like BERT BIBREF23 and BioBERT BIBREF24 also consider context but with deeper structures. Compared with those models, ELMo has less parameters and not easy to be overfitting. We trained our ELMo model on the MIMIC-III corpus. Since some sentences also appear in the test set, one may raise the concern of performance inflation in testing. However, we pre-trained ELMo using the whole corpus of MIMIC in an unsupervised way, so the classification labels were not visible during training. To train the ELMo model, we adapted code from AllenNLP, we set the word embedding dimension to 64 and applied two BiLSTM layers. For all of our experiments that involved with ELMo, we initialized the word embeddings from the last layer of ELMo.
Topic-attention We propose a neural topic-attention model for text classification. Our assumption is that the topic information plays an important role in doing the disambiguation given a sentence. For example, the abbreviation of the medical term FISH has two potential senses: general English fish as seafood and the sense of fluorescent in situ hybridization BIBREF17. The former case always goes with the topic of food, allergies while the other one appears in the topic of some examination test reports. In our model, a topic-attention module is applied to add topic information into the final sentence representation. This module calculates the distribution of topic-attention weights on a list of topic vectors. As shown in Figure SECREF5, we took the content vector $c_i$ (from soft-attention BIBREF22) and a topic matrix $T_{topic}=[t_1,t_2,..,t_j]$ (where each $t_i$ is a column vector in the figure and we illustrate four topics) as the inputs to the topic-attention module, and then calculated a weighted summation of the topic vectors. The final sentence representation $r_i$ for the sentence $i$ was calculated as the following:
where $W_{topic}$ and $b _ { topic }$ are trainable parameters, $\beta _{it}$ represents the topic-attention weights. The final sentence representation is denoted as $r_i$, and $[\cdot ,\cdot ]$ means concatenation. Here $s_i$ is the representation of the sentence which includes the topic information. The final sentence representation $r_i$ is the concatenation of $c_i$ and $s_i$, now we have both context-aware sentence representation and topic-related representation. Then we added a fully-connected layer, followed by a Softmax to do classification with cross-entropy loss function.
Topic Matrix To generate the topic matrix $T_{topic}$ as in Equation DISPLAY_FORM9, we propose a convolution-based method to generate topic vectors. We first pre-trained a topic model using the Latent Dirichlet Allocation (LDA) BIBREF25 model on MIMIC-III notes as used by the ELMo model. We set the number of topics to be 50 and ran for 30 iterations. Then we were able to get a list of top $k$ words (we set $k=100$) for each topic. To get the topic vector $t$ for a specific topic:
where $e_j$ (column vector) is the pre-trained Doc2vec word embedding for the top word $j$ from the current topic, and $Conv(\cdot )$ indicates a convolutional layer; $maxpool(\cdot )$ is a max pooling layer. We finally reshaped the output $t$ into a 1-dimension vector. Eventually we collected all topic vectors as the topic matrix $T_{topic}$ in Figure SECREF5.
Evaluation
We did three groups of experiments including two baseline groups and our proposed model. The first group was the TF-IDF features in Section SECREF3 for traditional models. The Naïve Bayesian classifier (NB) has the highest scores among all traditional methods. The second group of experiments used neural models described in Section SECREF3, where LSTM with self attention model (LSTM-self) has competitive results among this group, we choose this model as our base model. Notably, this is the first study that compares and evaluates LSTM-based models on the medical term abbreviation disambiguation task.
tableExperimental Results: we report macro F1 in all the experiments. figureTopic-attention Model
The last group contains the results of our proposed model with different settings. We used Topic Only setting on top of the base model, where we only added the topic-attention layer, and all the word embeddings were from our pre-trained Doc2vec model and were fine-tuned during back propagation. We could observe that compared with the base model, we have an improvement of 7.36% on accuracy and 9.69% on F1 score. Another model (ELMo Only) is to initialize the word embeddings of the base model using the pre-trained ELMo model, and here no topic information was added. Here we have higher scores than Topic Only, and the accuracy and F1 score increased by 9.87% and 12.26% respectively compared with the base model. We then conducted a combination of both(ELMo+Topic), where the word embeddings from the sentences were computed from the pre-trained ELMo model, and the topic representations were from the pre-trained Doc2vec model. We have a remarkable increase of 12.27% on the accuracy and 14.86% on F1 score.
To further compare our proposed topic-attention model and the base model, we report an average of area under the curve(AUC) score among all 30 terms: the base model has an average AUC of 0.7189, and our topic-attention model (ELMo+Topic) achieves an average AUC of 0.8196. We provide a case study in Appendix SECREF9. The results show that the model can benefit from ELMo, which considers contextual features, and the topic-attention module, which brings in topic information. We can conclude that under the few-shot learning setting, our proposed model can better capture the sentence features when only limited training samples are explored in a small-scale dataset.
Conclusion
In this paper, we propose a neural topic-attention model with few-shot learning for medical abbreviation disambiguation task. We also manually cleaned and collected training and testing data for this task, which we release to promote related research in NLP with clinical notes. In addition, we evaluated and compared a comprehensive set of baseline models, some of which had never been applied to the medical term abbreviation disambiguation task. Future work would be to adapt other models like BioBERT or BERT to our proposed topic-attention model. We will also extend the method into other clinical notes tasks such as drug name recognition and ICD-9 code auto-assigning BIBREF26. In addition, other LDA-based approach can be investigated.
Dataset Details
Figure FIGREF11 shows the histogram for the distribution of term-sense pair sample numbers. The X-axis gives the pair sample numbers and the Y-axis shows the counts. For example, the first bar shows that there are 43 term-sense pairs that have the sample number in the range of 0-50.
We also show histogram of class numbers for all terms in Figure FIGREF11. The Y-axis gives the counts while the X-axis gives the number of classes. For instance, the first bar means there are 12 terms contain 2 classes.
Testing Dataset
Since the training dataset is unbalanced and relatively small, it is hard to split it further into training and testing. Existing work BIBREF6, BIBREF5 conducted fold validation on the datasets and we found there are extreme rare senses for which only one or two samples exist. Besides, we believe that it is better to evaluate on a balanced dataset to determine whether it performs equally well on all classes. While most of the works deal with unbalanced training and testing that which may lead to very high accuracy if there is a dominating class in both training and testing set, the model may have a poor performance in rare testing cases because only a few samples have been seen during training. To be fair to all the classes, a good performance on these rare cases is also required otherwise it may lead to a severe situation. In this work, we are very interested to see how the model works among all senses especially the rare ones. Also, we can prevent the model from trivially predicting the dominant class and achieving high accuracy. As a result, we decided to create a dataset with the same number of samples for each case in the test dataset.
Our testing dataset takes MIMIC-III BIBREF18 and PubMed as data sources. Here we are referring to all the notes data from MIMIC-III (NOTEEVENTS table ) and Case Reports articles from PubMed, as these contents are close to medical notes. To create the test set, we first followed the approach by BIBREF6 who applied an auto-generating method. Initially, we built a term-sense dictionary from the training dataset. Then we did matching for the sense words or phrases in the MIMIC-III notes dataset, and once there is a match, we replaced the words or phrases with the abbreviation term . We then asked two researchers with a medical background to check the matched samples manually with the following judgment: given this sentence and the abbreviation term and its sense, do you think the content is enough for you to guess the sense and whether this is the right sense? To estimate the agreement on the annotations, we selected a subset which contains 120 samples randomly and let the two annotators annotate individually. We got a Kappa score BIBREF27 of 0.96, which is considered as a near perfect agreement. We then distributed the work to the two annotators, and each of them labeled a half of the dataset, which means each sample was only labeled by a single annotator. For some rare term-sense pairs, we failed to find samples from MIMIC-III. The annotators then searched these senses via PubMed data source manually, aiming to find clinical notes-like sentences. They picked good sentences from these results as testing samples where the keywords exist and the content is informative. For those senses that are extremely rare, we let the annotators create sentences in the clinical note style as testing samples according to their experiences. Eventually, we have a balanced testing dataset, where each term-sense pair has around 15 samples for testing (on average, each pair has 14.56 samples and the median sample number is 15), and a comparison with training dataset is shown in Figure FIGREF11. Due to the difficulty for collecting the testing dataset, we decided to only collect for a random selection of 30 terms. On average, it took few hours to generate testing samples for each abbreviation term per researcher.
Case Study
We now select two representative terms AC, IM and plot their receiver operating characteristic(ROC) curves. The term has a relatively large number of classes and the second one has extremely unbalanced training samples. We show the details in Table TABREF16. We have 8 classes in term AC. Figure FIGREF15(a) illustrates the results of our best performed model and Figure FIGREF15(b) shows the results of the base model. The accuracy and F1 score have an improvement from 0.3898 and 0.2830 to 0.4915 and 0.4059 respectively. Regarding the rare senses (for example, class 0, 1, 4 and 6), we can observe an increase in the ROC areas. Class 6 has an obvious improvement from 0.75 to 1.00. Such improvements in the rare senses make a huge difference in the reported average accuracy and F1 score, since we have a nearly equal number of samples for each class in the testing data. Similarly, we show the plots for IM term in Figure FIGREF15(c) and FIGREF15(d). IM has only two classes, but they are very unbalanced in training set, as shown in Table TABREF16. The accuracy and F1 scores improved from 0.6667 and 0.6250 to 0.8667 and 0.8667 respectively. We observe improvements in the ROC areas for both classes. This observation further shows that our model is more sensitive to all the class samples compared to the base model, even for the terms that have only a few samples in the training set. Again, by plotting the ROC curves and comparing AUC areas, we show that our model, which applies ELMo and topic-attention, has a better representation ability under the setting of few-shot learning.
Question:
To what baseline models is proposed model compared?
Answer:
| support vector machine classifier (SVM), logistic regression classifier (LR), Naive Bayes classifier (NB), random forest (RF), CNN, LSTM , LSTM-soft, LSTM-self | 328 | 18,599 | 18,601 | 18,662 | ... [The rest of the article is omitted]
|
3116453e35352a3a90ee5b12246dc7f2e60cfc59 | 3116453e35352a3a90ee5b12246dc7f2e60cfc59_1 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Medical text mining is an exciting area and is becoming attractive to natural language processing (NLP) researchers. Clinical notes are an example of text in the medical area that recent work has focused on BIBREF0, BIBREF1, BIBREF2. This work studies abbreviation disambiguation on clinical notes BIBREF3, BIBREF4, specifically those used commonly by physicians and nurses. Such clinical abbreviations can have a large number of meanings, depending on the specialty BIBREF5, BIBREF6. For example, the term MR can mean magnetic resonance, mitral regurgitation, mental retardation, medical record and the general English Mister (Mr.). Table TABREF1 illustrates such an example. Abbreviation disambiguation is an important task in medical text understanding task BIBREF7. Successful recognition of the abbreviations in the notes can contribute to downstream tasks such as classification, named entity recognition, and relation extraction BIBREF7.
Recent work focuses on formulating the abbreviation disambiguation task as a classification problem, where the possible senses of a given abbreviation term are pre-defined with the help of domain experts BIBREF6, BIBREF5. Traditional features such as part-of-speech (POS) and Term Frequency-Inverse Document Frequency (TF-IDF) are widely investigated for clinical notes classification. Classifiers like support vector machines (SVMs) and random forests (RFs) are used to make predictions BIBREF1. Such methods depend heavily on feature engineering. Recently, deep features have been investigated in the medical domain. Word embeddings BIBREF8 and Convolutional Neural Networks (CNNs) BIBREF9, BIBREF10 provide very competitive performance on text classification for clinical notes and abbreviation disambiguation task BIBREF0, BIBREF11, BIBREF12, BIBREF6. Beyond vanilla embeddings, BIBREF13 utilized contextual features to do abbreviation expansion. Another challenge is the difficulty in obtaining training data: clinical notes have many restrictions due to privacy issues and require domain experts to develop high-quality annotations, thus leading to limited annotated training and testing data. Another difficulty is that in the real world (and in the existing public datasets), some abbreviation term-sense pairs (for example, AB as abortion) have a very high frequency of occurrence BIBREF5, while others are rarely found. This long tail issue creates the challenge of training under unbalanced sample distributions. We tackle these problems in the setting of few-shot learning BIBREF14, BIBREF15 where only a few or a low number of samples can be found in the training dataset to make use of limited resources. We propose a model that combines deep contextual features based on ELMo BIBREF16 and topic information to solve the abbreviation disambiguation task.
Our contributions can be summarized as: 1) we re-examined and corrected an existing dataset for training and we collected a test set for evaluation with focus especially for rare senses; 2) we proposed a few-shot learning approach which combines topic information and contextualized word embeddings to solve clinical abbreviation disambiguation task. The implementation is available online; 3) as limited research are conducted on this particular task, we evaluated and compared a number of baseline methods including classical models and deep models comprehensively.
Datasets
Training Dataset UM Inventory BIBREF5 is a public dataset created by researchers from the University of Minnesota, containing about 37,500 training samples with 75 abbreviation terms. Existing work reports abbreviation disambiguation results on 50 abbreviation terms BIBREF6, BIBREF5, BIBREF17. However, after carefully reviewing this dataset, we found that it contains many samples where medical professionals disagree: wrong samples and uncategorized samples. Due to these mistakes and flaws of this dataset, we removed the erroneous samples and eventually selected 30 abbreviation terms as our training dataset that can be made public. Among all the abbreviation terms, we have 11,466 samples, and 93 term-sense pairs in total (on average 123.3 samples/term-sense pair and 382.2 samples/term). Some term-sense pairs are very popular with a larger number of training samples but some are not (typically less than 5), we call them rare-sense cases . More details can be found in Appendix SECREF7.
Testing Dataset Our testing dataset takes MIMIC-III BIBREF18 and PubMed as data sources. Here we are referring to all the notes data from MIMIC-III (NOTEEVENTS table ) and Case Reports articles from PubMed, as this type of contents are close to medical notes. We provide detailed information in Appendix SECREF8, including the steps to create the testing dataset. Eventually, we have a balanced testing dataset, where each term-sense pair has at least 11 and up to 15 samples for training (on average, each pair has 14.56 samples and the median sample number is 15).
Baselines
We conducted a comprehensive comparison with the baseline models, and some of them were never investigated for the abbreviation disambiguation task. We applied traditional features by simply taking the TF-IDF features as the inputs into the classic classifiers. Deep features are also considered: a Doc2vec model BIBREF19 was pre-trained using Gensim and these word embeddings were applied to initialize deep models and fine-tuned.
TF-IDF: We applied TF-IDF features as inputs to four classifiers: support vector machine classifier (SVM), logistic regression classifier (LR), Naive Bayes classifier (NB) and random forest (RF); CNN: We followed the same architecture as BIBREF9; LSTM: We applied an LSTM model BIBREF20 to classify the sentences with pre-trained word embeddings;LSTM-soft: We then added a soft-attention BIBREF21 layer on top of the LSTM model where we computed soft alignment scores over each of the hidden states; LSTM-self: We applied a self-attention layer BIBREF22 to LSTM model. We denote the content vector as $c_i$ for each sentence $i$, as the input to the last classification layer.
Topic-attention Model
ELMo ELMo is a new type of word representation introduced by BIBREF16 which considers contextual information captured by pre-trained BiLSTM language models. Similar works like BERT BIBREF23 and BioBERT BIBREF24 also consider context but with deeper structures. Compared with those models, ELMo has less parameters and not easy to be overfitting. We trained our ELMo model on the MIMIC-III corpus. Since some sentences also appear in the test set, one may raise the concern of performance inflation in testing. However, we pre-trained ELMo using the whole corpus of MIMIC in an unsupervised way, so the classification labels were not visible during training. To train the ELMo model, we adapted code from AllenNLP, we set the word embedding dimension to 64 and applied two BiLSTM layers. For all of our experiments that involved with ELMo, we initialized the word embeddings from the last layer of ELMo.
Topic-attention We propose a neural topic-attention model for text classification. Our assumption is that the topic information plays an important role in doing the disambiguation given a sentence. For example, the abbreviation of the medical term FISH has two potential senses: general English fish as seafood and the sense of fluorescent in situ hybridization BIBREF17. The former case always goes with the topic of food, allergies while the other one appears in the topic of some examination test reports. In our model, a topic-attention module is applied to add topic information into the final sentence representation. This module calculates the distribution of topic-attention weights on a list of topic vectors. As shown in Figure SECREF5, we took the content vector $c_i$ (from soft-attention BIBREF22) and a topic matrix $T_{topic}=[t_1,t_2,..,t_j]$ (where each $t_i$ is a column vector in the figure and we illustrate four topics) as the inputs to the topic-attention module, and then calculated a weighted summation of the topic vectors. The final sentence representation $r_i$ for the sentence $i$ was calculated as the following:
where $W_{topic}$ and $b _ { topic }$ are trainable parameters, $\beta _{it}$ represents the topic-attention weights. The final sentence representation is denoted as $r_i$, and $[\cdot ,\cdot ]$ means concatenation. Here $s_i$ is the representation of the sentence which includes the topic information. The final sentence representation $r_i$ is the concatenation of $c_i$ and $s_i$, now we have both context-aware sentence representation and topic-related representation. Then we added a fully-connected layer, followed by a Softmax to do classification with cross-entropy loss function.
Topic Matrix To generate the topic matrix $T_{topic}$ as in Equation DISPLAY_FORM9, we propose a convolution-based method to generate topic vectors. We first pre-trained a topic model using the Latent Dirichlet Allocation (LDA) BIBREF25 model on MIMIC-III notes as used by the ELMo model. We set the number of topics to be 50 and ran for 30 iterations. Then we were able to get a list of top $k$ words (we set $k=100$) for each topic. To get the topic vector $t$ for a specific topic:
where $e_j$ (column vector) is the pre-trained Doc2vec word embedding for the top word $j$ from the current topic, and $Conv(\cdot )$ indicates a convolutional layer; $maxpool(\cdot )$ is a max pooling layer. We finally reshaped the output $t$ into a 1-dimension vector. Eventually we collected all topic vectors as the topic matrix $T_{topic}$ in Figure SECREF5.
Evaluation
We did three groups of experiments including two baseline groups and our proposed model. The first group was the TF-IDF features in Section SECREF3 for traditional models. The Naïve Bayesian classifier (NB) has the highest scores among all traditional methods. The second group of experiments used neural models described in Section SECREF3, where LSTM with self attention model (LSTM-self) has competitive results among this group, we choose this model as our base model. Notably, this is the first study that compares and evaluates LSTM-based models on the medical term abbreviation disambiguation task.
tableExperimental Results: we report macro F1 in all the experiments. figureTopic-attention Model
The last group contains the results of our proposed model with different settings. We used Topic Only setting on top of the base model, where we only added the topic-attention layer, and all the word embeddings were from our pre-trained Doc2vec model and were fine-tuned during back propagation. We could observe that compared with the base model, we have an improvement of 7.36% on accuracy and 9.69% on F1 score. Another model (ELMo Only) is to initialize the word embeddings of the base model using the pre-trained ELMo model, and here no topic information was added. Here we have higher scores than Topic Only, and the accuracy and F1 score increased by 9.87% and 12.26% respectively compared with the base model. We then conducted a combination of both(ELMo+Topic), where the word embeddings from the sentences were computed from the pre-trained ELMo model, and the topic representations were from the pre-trained Doc2vec model. We have a remarkable increase of 12.27% on the accuracy and 14.86% on F1 score.
To further compare our proposed topic-attention model and the base model, we report an average of area under the curve(AUC) score among all 30 terms: the base model has an average AUC of 0.7189, and our topic-attention model (ELMo+Topic) achieves an average AUC of 0.8196. We provide a case study in Appendix SECREF9. The results show that the model can benefit from ELMo, which considers contextual features, and the topic-attention module, which brings in topic information. We can conclude that under the few-shot learning setting, our proposed model can better capture the sentence features when only limited training samples are explored in a small-scale dataset.
Conclusion
In this paper, we propose a neural topic-attention model with few-shot learning for medical abbreviation disambiguation task. We also manually cleaned and collected training and testing data for this task, which we release to promote related research in NLP with clinical notes. In addition, we evaluated and compared a comprehensive set of baseline models, some of which had never been applied to the medical term abbreviation disambiguation task. Future work would be to adapt other models like BioBERT or BERT to our proposed topic-attention model. We will also extend the method into other clinical notes tasks such as drug name recognition and ICD-9 code auto-assigning BIBREF26. In addition, other LDA-based approach can be investigated.
Dataset Details
Figure FIGREF11 shows the histogram for the distribution of term-sense pair sample numbers. The X-axis gives the pair sample numbers and the Y-axis shows the counts. For example, the first bar shows that there are 43 term-sense pairs that have the sample number in the range of 0-50.
We also show histogram of class numbers for all terms in Figure FIGREF11. The Y-axis gives the counts while the X-axis gives the number of classes. For instance, the first bar means there are 12 terms contain 2 classes.
Testing Dataset
Since the training dataset is unbalanced and relatively small, it is hard to split it further into training and testing. Existing work BIBREF6, BIBREF5 conducted fold validation on the datasets and we found there are extreme rare senses for which only one or two samples exist. Besides, we believe that it is better to evaluate on a balanced dataset to determine whether it performs equally well on all classes. While most of the works deal with unbalanced training and testing that which may lead to very high accuracy if there is a dominating class in both training and testing set, the model may have a poor performance in rare testing cases because only a few samples have been seen during training. To be fair to all the classes, a good performance on these rare cases is also required otherwise it may lead to a severe situation. In this work, we are very interested to see how the model works among all senses especially the rare ones. Also, we can prevent the model from trivially predicting the dominant class and achieving high accuracy. As a result, we decided to create a dataset with the same number of samples for each case in the test dataset.
Our testing dataset takes MIMIC-III BIBREF18 and PubMed as data sources. Here we are referring to all the notes data from MIMIC-III (NOTEEVENTS table ) and Case Reports articles from PubMed, as these contents are close to medical notes. To create the test set, we first followed the approach by BIBREF6 who applied an auto-generating method. Initially, we built a term-sense dictionary from the training dataset. Then we did matching for the sense words or phrases in the MIMIC-III notes dataset, and once there is a match, we replaced the words or phrases with the abbreviation term . We then asked two researchers with a medical background to check the matched samples manually with the following judgment: given this sentence and the abbreviation term and its sense, do you think the content is enough for you to guess the sense and whether this is the right sense? To estimate the agreement on the annotations, we selected a subset which contains 120 samples randomly and let the two annotators annotate individually. We got a Kappa score BIBREF27 of 0.96, which is considered as a near perfect agreement. We then distributed the work to the two annotators, and each of them labeled a half of the dataset, which means each sample was only labeled by a single annotator. For some rare term-sense pairs, we failed to find samples from MIMIC-III. The annotators then searched these senses via PubMed data source manually, aiming to find clinical notes-like sentences. They picked good sentences from these results as testing samples where the keywords exist and the content is informative. For those senses that are extremely rare, we let the annotators create sentences in the clinical note style as testing samples according to their experiences. Eventually, we have a balanced testing dataset, where each term-sense pair has around 15 samples for testing (on average, each pair has 14.56 samples and the median sample number is 15), and a comparison with training dataset is shown in Figure FIGREF11. Due to the difficulty for collecting the testing dataset, we decided to only collect for a random selection of 30 terms. On average, it took few hours to generate testing samples for each abbreviation term per researcher.
Case Study
We now select two representative terms AC, IM and plot their receiver operating characteristic(ROC) curves. The term has a relatively large number of classes and the second one has extremely unbalanced training samples. We show the details in Table TABREF16. We have 8 classes in term AC. Figure FIGREF15(a) illustrates the results of our best performed model and Figure FIGREF15(b) shows the results of the base model. The accuracy and F1 score have an improvement from 0.3898 and 0.2830 to 0.4915 and 0.4059 respectively. Regarding the rare senses (for example, class 0, 1, 4 and 6), we can observe an increase in the ROC areas. Class 6 has an obvious improvement from 0.75 to 1.00. Such improvements in the rare senses make a huge difference in the reported average accuracy and F1 score, since we have a nearly equal number of samples for each class in the testing data. Similarly, we show the plots for IM term in Figure FIGREF15(c) and FIGREF15(d). IM has only two classes, but they are very unbalanced in training set, as shown in Table TABREF16. The accuracy and F1 scores improved from 0.6667 and 0.6250 to 0.8667 and 0.8667 respectively. We observe improvements in the ROC areas for both classes. This observation further shows that our model is more sensitive to all the class samples compared to the base model, even for the terms that have only a few samples in the training set. Again, by plotting the ROC curves and comparing AUC areas, we show that our model, which applies ELMo and topic-attention, has a better representation ability under the setting of few-shot learning.
Question:
To what baseline models is proposed model compared?
Answer:
| support vector machine classifier (SVM), logistic regression classifier (LR), Naive Bayes classifier (NB), random forest (RF), CNN, LSTM , LSTM-soft, LSTM-self | 328 | 18,599 | 18,601 | 18,662 | ... [The rest of the article is omitted]
|
4e748cb2b5e74d905d9b24b53be6cfdf326e8054 | 4e748cb2b5e74d905d9b24b53be6cfdf326e8054_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Sarcasm is an intensive, indirect and complex construct that is often intended to express contempt or ridicule . Sarcasm, in speech, is multi-modal, involving tone, body-language and gestures along with linguistic artifacts used in speech. Sarcasm in text, on the other hand, is more restrictive when it comes to such non-linguistic modalities. This makes recognizing textual sarcasm more challenging for both humans and machines.
Sarcasm detection plays an indispensable role in applications like online review summarizers, dialog systems, recommendation systems and sentiment analyzers. This makes automatic detection of sarcasm an important problem. However, it has been quite difficult to solve such a problem with traditional NLP tools and techniques. This is apparent from the results reported by the survey from DBLP:journals/corr/JoshiBC16. The following discussion brings more insights into this.
Consider a scenario where an online reviewer gives a negative opinion about a movie through sarcasm: “This is the kind of movie you see because the theater has air conditioning”. It is difficult for an automatic sentiment analyzer to assign a rating to the movie and, in the absence of any other information, such a system may not be able to comprehend that prioritizing the air-conditioning facilities of the theater over the movie experience indicates a negative sentiment towards the movie. This gives an intuition to why, for sarcasm detection, it is necessary to go beyond textual analysis.
We aim to address this problem by exploiting the psycholinguistic side of sarcasm detection, using cognitive features extracted with the help of eye-tracking. A motivation to consider cognitive features comes from analyzing human eye-movement trajectories that supports the conjecture: Reading sarcastic texts induces distinctive eye movement patterns, compared to literal texts. The cognitive features, derived from human eye movement patterns observed during reading, include two primary feature types:
The cognitive features, along with textual features used in best available sarcasm detectors, are used to train binary classifiers against given sarcasm labels. Our experiments show significant improvement in classification accuracy over the state of the art, by performing such augmentation.
Related Work
Sarcasm, in general, has been the focus of research for quite some time. In one of the pioneering works jorgensen1984test explained how sarcasm arises when a figurative meaning is used opposite to the literal meaning of the utterance. In the word of clark1984pretense, sarcasm processing involves canceling the indirectly negated message and replacing it with the implicated one. giora1995irony, on the other hand, define sarcasm as a mode of indirect negation that requires processing of both negated and implicated messages. ivanko2003context define sarcasm as a six tuple entity consisting of a speaker, a listener, Context, Utterance, Literal Proposition and Intended Proposition and study the cognitive aspects of sarcasm processing.
Computational linguists have previously addressed this problem using rule based and statistical techniques, that make use of : (a) Unigrams and Pragmatic features BIBREF0 , BIBREF1 , BIBREF2 , BIBREF3 (b) Stylistic patterns BIBREF4 and patterns related to situational disparity BIBREF5 and (c) Hastag interpretations BIBREF6 , BIBREF7 .
Most of the previously done work on sarcasm detection uses distant supervision based techniques (ex: leveraging hashtags) and stylistic/pragmatic features (emoticons, laughter expressions such as “lol” etc). But, detecting sarcasm in linguistically well-formed structures, in absence of explicit cues or information (like emoticons), proves to be hard using such linguistic/stylistic features alone.
With the advent of sophisticated eye-trackers and electro/magneto-encephalographic (EEG/MEG) devices, it has been possible to delve deep into the cognitive underpinnings of sarcasm understanding. Filik2014, using a series of eye-tracking and EEG experiments try to show that for unfamiliar ironies, the literal interpretation would be computed first. They also show that a mismatch with context would lead to a re-interpretation of the statement, as being ironic. Camblin2007103 show that in multi-sentence passages, discourse congruence has robust effects on eye movements. This also implies that disrupted processing occurs for discourse incongruent words, even though they are perfectly congruous at the sentence level. In our previous work BIBREF8 , we augment cognitive features, derived from eye-movement patterns of readers, with textual features to detect whether a human reader has realized the presence of sarcasm in text or not.
The recent advancements in the literature discussed above, motivate us to explore gaze-based cognition for sarcasm detection. As far as we know, our work is the first of its kind.
Eye-tracking Database for Sarcasm Analysis
Sarcasm often emanates from incongruity BIBREF9 , which enforces the brain to reanalyze it BIBREF10 . This, in turn, affects the way eyes move through the text. Hence, distinctive eye-movement patterns may be observed in the case of successful processing of sarcasm in text in contrast to literal texts. This hypothesis forms the crux of our method for sarcasm detection and we validate this using our previously released freely available sarcasm dataset BIBREF8 enriched with gaze information.
Document Description
The database consists of 1,000 short texts, each having 10-40 words. Out of these, 350 are sarcastic and are collected as follows: (a) 103 sentences are from two popular sarcastic quote websites, (b) 76 sarcastic short movie reviews are manually extracted from the Amazon Movie Corpus BIBREF11 by two linguists. (c) 171 tweets are downloaded using the hashtag #sarcasm from Twitter. The 650 non-sarcastic texts are either downloaded from Twitter or extracted from the Amazon Movie Review corpus. The sentences do not contain words/phrases that are highly topic or culture specific. The tweets were normalized to make them linguistically well formed to avoid difficulty in interpreting social media lingo. Every sentence in our dataset carries positive or negative opinion about specific “aspects”. For example, the sentence “The movie is extremely well cast” has positive sentiment about the aspect “cast”.
The annotators were seven graduate students with science and engineering background, and possess good English proficiency. They were given a set of instructions beforehand and are advised to seek clarifications before they proceed. The instructions mention the nature of the task, annotation input method, and necessity of head movement minimization during the experiment.
Task Description
The task assigned to annotators was to read sentences one at a time and label them with with binary labels indicating the polarity (i.e., positive/negative). Note that, the participants were not instructed to annotate whether a sentence is sarcastic or not., to rule out the Priming Effect (i.e., if sarcasm is expected beforehand, processing incongruity becomes relatively easier BIBREF12 ). The setup ensures its “ecological validity” in two ways: (1) Readers are not given any clue that they have to treat sarcasm with special attention. This is done by setting the task to polarity annotation (instead of sarcasm detection). (2) Sarcastic sentences are mixed with non sarcastic text, which does not give prior knowledge about whether the forthcoming text will be sarcastic or not.
The eye-tracking experiment is conducted by following the standard norms in eye-movement research BIBREF13 . At a time, one sentence is displayed to the reader along with the “aspect” with respect to which the annotation has to be provided. While reading, an SR-Research Eyelink-1000 eye-tracker (monocular remote mode, sampling rate 500Hz) records several eye-movement parameters like fixations (a long stay of gaze) and saccade (quick jumping of gaze between two positions of rest) and pupil size.
The accuracy of polarity annotation varies between 72%-91% for sarcastic texts and 75%-91% for non-sarcastic text, showing the inherent difficulty of sentiment annotation, when sarcasm is present in the text under consideration. Annotation errors may be attributed to: (a) lack of patience/attention while reading, (b) issues related to text comprehension, and (c) confusion/indecisiveness caused due to lack of context.
For our analysis, we do not discard the incorrect annotations present in the database. Since our system eventually aims to involve online readers for sarcasm detection, it will be hard to segregate readers who misinterpret the text. We make a rational assumption that, for a particular text, most of the readers, from a fairly large population, will be able to identify sarcasm. Under this assumption, the eye-movement parameters, averaged across all readers in our setting, may not be significantly distorted by a few readers who would have failed to identify sarcasm. This assumption is applicable for both regular and multi-instance based classifiers explained in section SECREF6 .
Analysis of Eye-movement Data
We observe distinct behavior during sarcasm reading, by analyzing the “fixation duration on the text” (also referred to as “dwell time” in the literature) and “scanpaths” of the readers.
Variation in the Average Fixation Duration per Word
Since sarcasm in text can be expected to induce cognitive load, it is reasonable to believe that it would require more processing time BIBREF14 . Hence, fixation duration normalized over total word count should usually be higher for a sarcastic text than for a non-sarcastic one. We observe this for all participants in our dataset, with the average fixation duration per word for sarcastic texts being at least 1.5 times more than that of non-sarcastic texts. To test the statistical significance, we conduct a two-tailed t-test (assuming unequal variance) to compare the average fixation duration per word for sarcastic and non-sarcastic texts. The hypothesized mean difference is set to 0 and the error tolerance limit ( INLINEFORM0 ) is set to 0.05. The t-test analysis, presented in Table TABREF11 , shows that for all participants, a statistically significant difference exists between the average fixation duration per word for sarcasm (higher average fixation duration) and non-sarcasm (lower average fixation duration). This affirms that the presence of sarcasm affects the duration of fixation on words.
It is important to note that longer fixations may also be caused by other linguistic subtleties (such as difficult words, ambiguity and syntactically complex structures) causing delay in comprehension, or occulomotor control problems forcing readers to spend time adjusting eye-muscles. So, an elevated average fixation duration per word may not sufficiently indicate the presence of sarcasm. But we would also like to share that, for our dataset, when we considered readability (Flesch readability ease-score BIBREF15 ), number of words in a sentence and average character per word along with the sarcasm label as the predictors of average fixation duration following a linear mixed effect model BIBREF16 , sarcasm label turned out to be the most significant predictor with a maximum slope. This indicates that average fixation duration per word has a strong connection with the text being sarcastic, at least in our dataset.
We now analyze scanpaths to gain more insights into the sarcasm comprehension process.
Analysis of Scanpaths
Scanpaths are line-graphs that contain fixations as nodes and saccades as edges; the radii of the nodes represent the fixation duration. A scanpath corresponds to a participant's eye-movement pattern while reading a particular sentence. Figure FIGREF14 presents scanpaths of three participants for the sarcastic sentence S1 and the non-sarcastic sentence S2. The x-axis of the graph represents the sequence of words a reader reads, and the y-axis represents a temporal sequence in milliseconds.
Consider a sarcastic text containing incongruous phrases A and B. Our qualitative scanpath-analysis reveals that scanpaths with respect to sarcasm processing have two typical characteristics. Often, a long regression - a saccade that goes to a previously visited segment - is observed when a reader starts reading B after skimming through A. In a few cases, the fixation duration on A and B are significantly higher than the average fixation duration per word. In sentence S1, we see long and multiple regressions from the two incongruous phrases “misconception” and “cherish”, and a few instances where phrases “always cherish” and “original misconception” are fixated longer than usual. Such eye-movement behaviors are not seen for S2.
Though sarcasm induces distinctive scanpaths like the ones depicted in Figure FIGREF14 in the observed examples, presence of such patterns is not sufficient to guarantee sarcasm; such patterns may also possibly arise from literal texts. We believe that a combination of linguistic features, readability of text and features derived from scanpaths would help discriminative machine learning models learn sarcasm better.
Features for Sarcasm Detection
We describe the features used for sarcasm detection in Table . The features enlisted under lexical,implicit incongruity and explicit incongruity are borrowed from various literature (predominantly from joshi2015harnessing). These features are essential to separate sarcasm from other forms semantic incongruity in text (for example ambiguity arising from semantic ambiguity or from metaphors). Two additional textual features viz. readability and word count of the text are also taken under consideration. These features are used to reduce the effect of text hardness and text length on the eye-movement patterns.
Simple Gaze Based Features
Readers' eye-movement behavior, characterized by fixations, forward saccades, skips and regressions, can be directly quantified by simple statistical aggregation (i.e., either computing features for individual participants and then averaging or performing a multi-instance based learning as explained in section SECREF6 ). Since these eye-movement attributes relate to the cognitive process in reading BIBREF17 , we consider these as features in our model. Some of these features have been reported by sarcasmunderstandability for modeling sarcasm understandability of readers. However, as far as we know, these features are being introduced in NLP tasks like textual sarcasm detection for the first time. The values of these features are believed to increase with the increase in the degree of surprisal caused by incongruity in text (except skip count, which will decrease).
Complex Gaze Based Features
For these features, we rely on a graph structure, namely “saliency graphs", derived from eye-gaze information and word sequences in the text.
For each reader and each sentence, we construct a “saliency graph”, representing the reader's attention characteristics. A saliency graph for a sentence INLINEFORM0 for a reader INLINEFORM1 , represented as INLINEFORM2 , is a graph with vertices ( INLINEFORM3 ) and edges ( INLINEFORM4 ) where each vertex INLINEFORM5 corresponds to a word in INLINEFORM6 (may not be unique) and there exists an edge INLINEFORM7 between vertices INLINEFORM8 and INLINEFORM9 if R performs at least one saccade between the words corresponding to INLINEFORM10 and INLINEFORM11 .
Figure FIGREF15 shows an example of a saliency graph.A saliency graph may be weighted, but not necessarily connected, for a given text (as there may be words in the given text with no fixation on them). The “complex” gaze features derived from saliency graphs are also motivated by the theory of incongruity. For instance, Edge Density of a saliency graph increases with the number of distinct saccades, which could arise from the complexity caused by presence of sarcasm. Similarly, the highest weighted degree of a graph is expected to be higher, if the node corresponds to a phrase, incongruous to some other phrase in the text.
The Sarcasm Classifier
We interpret sarcasm detection as a binary classification problem. The training data constitutes 994 examples created using our eye-movement database for sarcasm detection. To check the effectiveness of our feature set, we observe the performance of multiple classification techniques on our dataset through a stratified 10-fold cross validation. We also compare the classification accuracy of our system and the best available systems proposed by riloff2013sarcasm and joshi2015harnessing on our dataset. Using Weka BIBREF18 and LibSVM BIBREF19 APIs, we implement the following classifiers:
Results
Table TABREF17 shows the classification results considering various feature combinations for different classifiers and other systems. These are:
Unigram (with principal components of unigram feature vectors),
Sarcasm (the feature-set reported by joshi2015harnessing subsuming unigram features and features from other reported systems)
Gaze (the simple and complex cognitive features we introduce, along with readability and word count features), and
Gaze+Sarcasm (the complete set of features).
For all regular classifiers, the gaze features are averaged across participants and augmented with linguistic and sarcasm related features. For the MILR classifier, the gaze features derived from each participant are augmented with linguistic features and thus, a multi instance “bag” of features is formed for each sentence in the training data. This multi-instance dataset is given to an MILR classifier, which follows the standard multi instance assumption to derive class-labels for each bag.
For all the classifiers, our feature combination outperforms the baselines (considering only unigram features) as well as BIBREF3 , with the MILR classifier getting an F-score improvement of 3.7% and Kappa difference of 0.08. We also achieve an improvement of 2% over the baseline, using SVM classifier, when we employ our feature set. We also observe that the gaze features alone, also capture the differences between sarcasm and non-sarcasm classes with a high-precision but a low recall.
To see if the improvement obtained is statistically significant over the state-of-the art system with textual sarcasm features alone, we perform McNemar test. The output of the SVM classifier using only linguistic features used for sarcasm detection by joshi2015harnessing and the output of the MILR classifier with the complete set of features are compared, setting threshold INLINEFORM0 . There was a significant difference in the classifier's accuracy with p(two-tailed) = 0.02 with an odds-ratio of 1.43, showing that the classification accuracy improvement is unlikely to be observed by chance in 95% confidence interval.
Considering Reading Time as a Cognitive Feature along with Sarcasm Features
One may argue that, considering simple measures of reading effort like “reading time” as cognitive feature instead of the expensive eye-tracking features for sarcasm detection may be a cost-effective solution. To examine this, we repeated our experiments with “reading time” considered as the only cognitive feature, augmented with the textual features. The F-scores of all the classifiers turn out to be close to that of the classifiers considering sarcasm feature alone and the difference in the improvement is not statistically significant ( INLINEFORM0 ). One the other hand, F-scores with gaze features are superior to the F-scores when reading time is considered as a cognitive feature.
How Effective are the Cognitive Features
We examine the effectiveness of cognitive features on the classification accuracy by varying the input training data size. To examine this, we create a stratified (keeping the class ratio constant) random train-test split of 80%:20%. We train our classifier with 100%, 90%, 80% and 70% of the training data with our whole feature set, and the feature combination from joshi2015harnessing. The goodness of our system is demonstrated by improvements in F-score and Kappa statistics, shown in Figure FIGREF22 .
We further analyze the importance of features by ranking the features based on (a) Chi squared test, and (b) Information Gain test, using Weka's attribute selection module. Figure FIGREF23 shows the top 20 ranked features produced by both the tests. For both the cases, we observe 16 out of top 20 features to be gaze features. Further, in each of the cases, Average Fixation Duration per Word and Largest Regression Position are seen to be the two most significant features.
Example Cases
Table TABREF21 shows a few example cases from the experiment with stratified 80%-20% train-test split.
Example sentence 1 is sarcastic, and requires extra-linguistic knowledge (about poor living conditions at Manchester). Hence, the sarcasm detector relying only on textual features is unable to detect the underlying incongruity. However, our system predicts the label successfully, possibly helped by the gaze features.
Similarly, for sentence 2, the false sense of presence of incongruity (due to phrases like “Helped me” and “Can't stop”) affects the system with only linguistic features. Our system, though, performs well in this case also.
Sentence 3 presents a false-negative case where it was hard for even humans to get the sarcasm. This is why our gaze features (and subsequently the complete set of features) account for erroneous prediction.
In sentence 4, gaze features alone false-indicate presence of incongruity, whereas the system predicts correctly when gaze and linguistic features are taken together.
From these examples, it can be inferred that, only gaze features would not have sufficed to rule out the possibility of detecting other forms of incongruity that do not result in sarcasm.
Error Analysis
Errors committed by our system arise from multiple factors, starting from limitations of the eye-tracker hardware to errors committed by linguistic tools and resources. Also, aggregating various eye-tracking parameters to extract the cognitive features may have caused information loss in the regular classification setting.
Conclusion
In the current work, we created a novel framework to detect sarcasm, that derives insights from human cognition, that manifests over eye movement patterns. We hypothesized that distinctive eye-movement patterns, associated with reading sarcastic text, enables improved detection of sarcasm. We augmented traditional linguistic features with cognitive features obtained from readers' eye-movement data in the form of simple gaze-based features and complex features derived from a graph structure. This extended feature-set improved the success rate of the sarcasm detector by 3.7%, over the best available system. Using cognitive features in an NLP Processing system like ours is the first proposal of its kind.
Our general approach may be useful in other NLP sub-areas like sentiment and emotion analysis, text summarization and question answering, where considering textual clues alone does not prove to be sufficient. We propose to augment this work in future by exploring deeper graph and gaze features. We also propose to develop models for the purpose of learning complex gaze feature representation, that accounts for the power of individual eye movement patterns along with the aggregated patterns of eye movements.
Acknowledgments
We thank the members of CFILT Lab, especially Jaya Jha and Meghna Singh, and the students of IIT Bombay for their help and support.
Question:
What traditional linguistics features did they use?
Answer:
| unanswerable | 328 | 23,949 | 23,951 | 24,012 | ... [The rest of the article is omitted]
|
b970f48d30775d3468952795bc72976baab3438e | b970f48d30775d3468952795bc72976baab3438e_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
In June 2015, the operators of the online discussion site Reddit banned several communities under new anti-harassment rules. BIBREF0 used this opportunity to combine rich online data with computational methods to study a current question: Does eliminating these “echo chambers” diminish the amount of hate speech overall? Exciting opportunities like these, at the intersection of “thick” cultural and societal questions on the one hand, and the computational analysis of rich textual data on larger-than-human scales on the other, are becoming increasingly common.
Indeed, computational analysis is opening new possibilities for exploring challenging questions at the heart of some of the most pressing contemporary cultural and social issues. While a human reader is better equipped to make logical inferences, resolve ambiguities, and apply cultural knowledge than a computer, human time and attention are limited. Moreover, many patterns are not obvious in any specific context, but only stand out in the aggregate. For example, in a landmark study, BIBREF1 analyzed the authorship of The Federalist Papers using a statistical text analysis by focusing on style, based on the distribution of function words, rather than content. As another example, BIBREF2 studied what defines English haiku and showed how computational analysis and close reading can complement each other. Computational approaches are valuable precisely because they help us identify patterns that would not otherwise be discernible.
Yet these approaches are not a panacea. Examining thick social and cultural questions using computational text analysis carries significant challenges. For one, texts are culturally and socially situated. They reflect the ideas, values and beliefs of both their authors and their target audiences, and such subtleties of meaning and interpretation are difficult to incorporate in computational approaches. For another, many of the social and cultural concepts we seek to examine are highly contested — hate speech is just one such example. Choices regarding how to operationalize and analyze these concepts can raise serious concerns about conceptual validity and may lead to shallow or obvious conclusions, rather than findings that reflect the depth of the questions we seek to address.
These are just a small sample of the many opportunities and challenges faced in computational analyses of textual data. New possibilities and frustrating obstacles emerge at every stage of research, from identification of the research question to interpretation of the results. In this article, we take the reader through a typical research process that involves measuring social or cultural concepts using computational methods, discussing both the opportunities and complications that often arise. In the Reddit case, for example, hate speech is measured, however imperfectly, by the presence of particular words semi-automatically extracted from a machine learning algorithm. Operationalizations are never perfect translations, and are often refined over the course of an investigation, but they are crucial.
We begin our exploration with the identification of research questions, proceed through data selection, conceptualization, and operationalization, and end with analysis and the interpretation of results. The research process sounds more or less linear this way, but each of these phases overlaps, and in some instances turns back upon itself. The analysis phase, for example, often feeds back into the original research questions, which may continue to evolve for much of the project. At each stage, our discussion is critically informed by insights from the humanities and social sciences, fields that have focused on, and worked to tackle, the challenges of textual analysis—albeit at smaller scales—since their inception.
In describing our experiences with computational text analysis, we hope to achieve three primary goals. First, we aim to shed light on thorny issues not always at the forefront of discussions about computational text analysis methods. Second, we hope to provide a set of best practices for working with thick social and cultural concepts. Our guidance is based on our own experiences and is therefore inherently imperfect. Still, given our diversity of disciplinary backgrounds and research practices, we hope to capture a range of ideas and identify commonalities that will resonate for many. And this leads to our final goal: to help promote interdisciplinary collaborations. Interdisciplinary insights and partnerships are essential for realizing the full potential of any computational text analysis that involves social and cultural concepts, and the more we are able to bridge these divides, the more fruitful we believe our work will be.
Research questions
We typically start by identifying the questions we wish to explore. Can text analysis provide a new perspective on a “big question” that has been attracting interest for years? Or can we raise new questions that have only recently emerged, for example about social media? For social scientists working in computational analysis, the questions are often grounded in theory, asking: How can we explain what we observe? These questions are also influenced by the availability and accessibility of data sources. For example, the choice to work with data from a particular social media platform may be partly determined by the fact that it is freely available, and this will in turn shape the kinds of questions that can be asked. A key output of this phase are the concepts to measure, for example: influence; copying and reproduction; the creation of patterns of language use; hate speech. Computational analysis of text motivated by these questions is insight driven: we aim to describe a phenomenon or explain how it came about. For example, what can we learn about how and why hate speech is used or how this changes over time? Is hate speech one thing, or does it comprise multiple forms of expression? Is there a clear boundary between hate speech and other types of speech, and what features make it more or less ambiguous? In these cases, it is critical to communicate high-level patterns in terms that are recognizable.
This contrasts with much of the work in computational text analysis, which tends to focus on automating tasks that humans perform inefficiently. These tasks range from core linguistically motivated tasks that constitute the backbone of natural language processing, such as part-of-speech tagging and parsing, to filtering spam and detecting sentiment. Many tasks are motivated by applications, for example to automatically block online trolls. Success, then, is often measured by performance, and communicating why a certain prediction was made—for example, why a document was labeled as positive sentiment, or why a word was classified as a noun—is less important than the accuracy of the prediction itself. The approaches we use and what we mean by `success' are thus guided by our research questions.
Domain experts and fellow researchers can provide feedback on questions and help with dynamically revising them. For example, they may say “we already think we know that”, “that's too naïve”, “that doesn't reflect social reality” (negative); “two major camps in the field would give different answers to that question” (neutral); “we tried to look at that back in the 1960s, but we didn't have the technology” (positive); and “that sounds like something that people who made that archive would love”, “that's a really fundamental question” (very positive).
Sometimes we also hope to connect to multiple disciplines. For example, while focusing on the humanistic concerns of an archive, we could also ask social questions such as “is this archive more about collaborative processes, culture-building or norm creation?” or “how well does this archive reflect the society in which it is embedded?" BIBREF3 used quantitative methods to tell a story about Darwin's intellectual development—an essential biographical question for a key figure in the history of science. At the same time, their methods connected Darwin's development to the changing landscape of Victorian scientific culture, allowing them to contrast Darwin's “foraging” in the scientific literature of his time to the ways in which that literature was itself produced. Finally, their methods provided a case study, and validation of technical approaches, for cognitive scientists who are interested in how people explore and exploit sources of knowledge.
Questions about potential “dual use” may also arise. Returning to our introductory example, BIBREF0 started with a deceptively simple question: if an internet platform eliminates forums for hate speech, does this impact hate speech in other forums? The research was motivated by the belief that a rising tide of online hate speech was (and is) making the internet increasingly unfriendly for disempowered groups, including minorities, women, and LBGTQ individuals. Yet the possibility of dual use troubled the researchers from the onset. Could the methodology be adopted to target the speech of groups like Black Lives Matter? Could it be adopted by repressive governments to minimize online dissent? While these concerns remained, they concluded that hypothetical dual use scenarios did not outweigh the tangible contribution this research could offer towards making the online environment more equal and just.
Data
The next step involves deciding on the data sources, collecting and compiling the dataset, and inspecting its metadata.
Data acquisition
Many scholars in the humanities and the social sciences work with sources that are not available in digital form, and indeed may never be digitized. Others work with both analogue and digitized materials, and the increasing digitization of archives has opened opportunities to study these archives in new ways. We can go to the canonical archive or open up something that nobody has studied before. For example, we might focus on major historical moments (French Revolution, post-Milosevic Serbia) or critical epochs (Britain entering the Victorian era, the transition from Latin to proto-Romance). Or, we could look for records of how people conducted science, wrote and consumed literature, and worked out their philosophies.
A growing number of researchers work with born-digital sources or data. Born-digital data, e.g., from social media, generally do not involve direct elicitation from participants and therefore enable unobtrusive measurements BIBREF5 , BIBREF6 . In contrast, methods like surveys sometimes elicit altered responses from participants, who might adapt their responses to what they think is expected. Moreover, born-digital data is often massive, enabling large-scale studies of language and behavior in a variety of social contexts.
Still, many scholars in the social sciences and humanities work with multiple data sources. The variety of sources typically used means that more than one data collection method is often required. For example, a project examining coverage of a UK General Election, could draw data from traditional media, web archives, Twitter and Facebook, campaign manifestos, etc. and might combine textual analysis of these materials with surveys, laboratory experiments, or field observations offline. In contrast, many computational studies based on born-digital data have focused on one specific source, such as Twitter.
The use of born-digital data raises ethical concerns. Although early studies often treated privacy as a binary construct, many now acknowledge its complexity BIBREF7 . Conversations on private matters can be posted online, visible for all, but social norms regarding what should be considered public information may differ from the data's explicit visibility settings. Often no informed consent has been obtained, raising concerns and challenges regarding publishing content and potentially harmful secondary uses BIBREF8 , BIBREF4 .
Recently, concerns about potential harms stemming from secondary uses have led a number of digital service providers to restrict access to born-digital data. Facebook and Twitter, for example, have reduced or eliminated public access to their application programming interfaces (APIs) and expressed hesitation about allowing academic researchers to use data from their platforms to examine certain sensitive or controversial topics. Despite the seeming abundance of born-digital data, we therefore cannot take its availability for granted.
Working with data that someone else has acquired presents additional problems related to provenance and contextualisation. It may not always be possible to determine the criteria applied during the creation process. For example, why were certain newspapers digitized but not others, and what does this say about the collection? Similar questions arise with the use of born-digital data. For instance, when using the Internet Archive’s Wayback Machine to gather data from archived web pages, we need to consider what pages were captured, which are likely missing, and why.
We must often repurpose born-digital data (e.g., Twitter was not designed to measure public opinion), but data biases may lead to spurious results and limit justification for generalization. In particular, data collected via black box APIs designed for commercial, not research, purposes are likely to introduce biases into the inferences we draw, and the closed nature of these APIs means we rarely know what biases are introduced, let alone how severely they might impact our research BIBREF10 . These, however, are not new problems. Historians, for example, have always understood that their sources were produced within particular contexts and for particular purposes, which are not always apparent to us.
Non-representative data can still be useful for making comparisons within a sample. In the introductory example on hate speech BIBREF0 , the Reddit forums do not present a comprehensive or balanced picture of hate speech: the writing is almost exclusively in English, the targets of hate speech are mainly restricted (e.g., to black people, or women), and the population of writers is shaped by Reddit's demographics, which skew towards young white men. These biases limit the generalizability of the findings, which cannot be extrapolated to other languages, other types of hate speech, and other demographic groups. However, because the findings are based on measurements on the same sort of hate speech and the same population of writers, as long as the collected data are representative of this specific population, these biases do not pose an intractable validity problem if claims are properly restricted.
The size of many newly available datasets is one of their most appealing characteristics. Bigger datasets often make statistics more robust. The size needed for a computational text analysis depends on the research goal: When it involves studying rare events, bigger datasets are needed. However, larger is not always better. Some very large archives are “secretly” collections of multiple and distinct processes that no in-field scholar would consider related. For example, Google Books is frequently used to study cultural patterns, but the over-representation of scientific articles in Google books can be problematic BIBREF11 . Even very large born-digital datasets usually cover limited timespans compared to, e.g., the Gutenberg archive of British novels.
This stage of the research also raises important questions about fairness. Are marginalized groups, for example, represented in the tweets we have collected? If not, what types of biases might result from analyses relying on those tweets?
Local experts and “informants” can help navigate the data. They can help understand the role an archive plays in the time and place. They might tell us: Is this the central archive, or a peripheral one? What makes it unusual? Or they might tell us how certain underrepresented communities use a social media platform and advise us on strategies for ensuring our data collection includes their perspectives.
However, when it is practically infeasible to navigate the data in this way—for instance, when we cannot determine what is missing from Twitter's Streaming API or what webpages are left out of the Internet Archive—we should be open about the limitations of our analyses, acknowledging the flaws in our data and drawing cautious and reasonable conclusions from them. In all cases, we should report the choices we have made when creating or re-using any dataset.
Compiling data
After identifying the data source(s), the next step is compiling the data. This step is fundamental: if the sources cannot support a convincing result, no result will be convincing. In many cases, this involves defining a “core" set of documents and a “comparison" set. We often have a specific set of documents in mind: an author's work, a particular journal, a time period. But if we want to say that this “core" set has some distinctive property, we need a “comparison" set. Expanding the collection beyond the documents that we would immediately think of has the beneficial effect of increasing our sample size. Having more sources increases the chance that we will notice something consistent across many individually varying contexts.
Comparing sets of documents can sometimes support causal inference, presented as a contrast between a treatment group and a control. In BIBREF0 , the treatment consisted of the text written in the two forums that were eventually closed by Reddit. However, identifying a control group required a considerable amount of time and effort. Reddit is a diverse platform, with a wide variety of interactional and linguistic styles; it would be pointless to compare hate speech forums against forums dedicated to, say, pictures of wrecked bicycles. Chandrasekharan et al. used a matching design, populating the control group with forums that were as similar as possible to the treatment group, but were not banned from Reddit. The goal is to estimate the counterfactual scenario: in this case, what would have happened had the site not taken action against these specific forums? An ideal control would make it possible to distinguish the effect of the treatment — closing the forums — from other idiosyncratic properties of texts that were treated.
We also look for categories of documents that might not be useful. We might remove documents that are meta-discourse, like introductions and notes, or documents that are in a language that is not the primary language of the collection, or duplicates when we are working with archived web pages. However, we need to carefully consider the potential consequences of information we remove. Does its removal alter the data, or the interpretation of the data, we are analyzing? Are we losing anything that might be valuable at a later stage?
Labels and metadata
Sometimes all we have is documents, but often we want to look at documents in the context of some additional information, or metadata. This additional information could tell us about the creation of documents (date, author, forum), or about the reception of documents (flagged as hate speech, helpful review). Information about text segments can be extremely valuable, but it is also prone to errors, inconsistencies, bias, and missing information. Examining metadata is a good way to check a collection's balance and representativeness. Are sources disproportionately of one form? Is the collection missing a specific time window? This type of curation can be extremely time consuming as it may require expert labeling, but it often leads to the most compelling results. Sometimes metadata are also used as target labels to develop machine learning models. But using them as a “ground truth” requires caution. Labels sometimes mean something different than we expect. For example, a down vote for a social media post could indicate that the content is offensive, or that the voter simply disagreed with the expressed view.
Conceptualization
A core step in many analyses is translating social and cultural concepts (such as hate speech, rumor, or conversion) into measurable quantities. Before we can develop measurements for these concepts (the operationalization step, or the “implementation” step as denoted by BIBREF12 ), we need to define them. In the conceptualization phase we often start with questions such as: who are the domain experts, and how have they approached the topic? We are looking for a definition of the concept that is flexible enough to apply on our dataset, yet formal enough for computational research. For example, our introductory study on hate speech BIBREF0 used a statement on hate speech produced by the European Union Court of Human Rights. The goal was not to implement this definition directly in software but to use it as a reference point to anchor subsequent analyses.
If we want to move beyond the use of ad hoc definitions, it can be useful to distinguish between what political scientists Adcock and Collier call the “background concept” and the “systematized concept” BIBREF13 . The background concept comprises the full and diverse set of meanings that might be associated with a particular term. This involves delving into theoretical, conceptual, and empirical studies to assess how a concept has been defined by other scholars and, most importantly, to determine which definition is most appropriate for the particular research question and the theoretical framework in which it is situated. That definition, in turn, represents the systematized concept: the formulation that is adopted for the study.
It is important to consider that for social and cultural concepts there is no absolute ground truth. There are often multiple valid definitions for a concept (the “background” concept in the terms of Adcock and Collier), and definitions might be contested over time. This may be uncomfortable for computer scientists, whose primary measure of success is often based on comparing a model's output against “ground truth” or a “gold standard”, e.g., by comparing a sentiment classifier's output against manual annotations. However, the notion of ground truth is uncommon in the humanities and the social sciences and it is often taken too far in machine learning. BIBREF14 notes that in literary criticism and the digital humanities more broadly “interpretation, ambiguity, and argumentation are prized far above ground truth and definitive conclusions". BIBREF15 draw attention to the different attitudes of literary scholars and computational linguists towards ambiguity, stating that “In Computational Linguistics [..] ambiguity is almost uniformly treated as a problem to be solved; the focus is on disambiguation, with the assumption that one true, correct interpretation exists." The latter is probably true for tasks such as spam filtering, but in the social sciences and the humanities many relevant concepts are fundamentally unobservable, such as latent traits of political actors BIBREF16 or cultural fit in organizations BIBREF17 , leading to validation challenges. Moreover, when the ground truth comes from people, it may be influenced by ideological priors, priming, simple differences of opinion or perspective, and many other factors BIBREF18 . We return to this issue in our discussions on validation and analysis.
Operationalization
In this phase we develop measures (or, “operationalizations”, or “indicators”) for the concepts of interest, a process called “operationalization”. Regardless of whether we are working with computers, the output produced coincides with Adcock and Collier's “scores”—the concrete translation and output of the systematized concept into numbers or labels BIBREF13 . Choices made during this phase are always tied to the question “Are we measuring what we intend to measure?” Does our operationalization match our conceptual definition? To ensure validity we must recognize gaps between what is important and what is easy to measure. We first discuss modeling considerations. Next, we describe several frequently used computational approaches and their limitations and strengths.
Modeling considerations
The variables (both predictors and outcomes) are rarely simply binary or categorical. For example, a study on language use and age could focus on chronological age (instead of, e.g., social age BIBREF19 ). However, even then, age can be modeled in different ways. Discretization can make the modeling easier and various NLP studies have modeled age as a categorical variable BIBREF20 . But any discretization raises questions: How many categories? Where to place the boundaries? Fine distinctions might not always be meaningful for the analysis we are interested in, but categories that are too broad can threaten validity. Other interesting variables include time, space, and even the social network position of the author. It is often preferable to keep the variable in its most precise form. For example, BIBREF21 perform exploration in the context of hypothesis testing by using latitude and longitude coordinates — the original metadata attached to geotagged social media such as tweets — rather than aggregating into administrative units such as counties or cities. This is necessary when such administrative units are unlikely to be related to the target concept, as is the case in their analysis of dialect differences. Focusing on precise geographical coordinates also makes it possible to recognize fine-grained effects, such as language variation across the geography of a city.
Using a particular classification scheme means deciding which variations are visible, and which ones are hidden BIBREF22 . We are looking for a categorization scheme for which it is feasible to collect a large enough labeled document collection (e.g., to train supervised models), but which is also fine-grained enough for our purposes. Classification schemes rarely exhibit the ideal properties, i.e., that they are consistent, their categories are mutually exclusive, and that the system is complete BIBREF22 . Borderline cases are challenging, especially with social and cultural concepts, where the boundaries are often not clear-cut. The choice of scheme can also have ethical implications BIBREF22 . For example, gender is usually represented as a binary variable in NLP and computational models tend to learn gender-stereotypical patterns. The operationalization of gender in NLP has been challenged only recently BIBREF23 , BIBREF24 , BIBREF25 .
Supervised and unsupervised learning are the most common approaches to learning from data. With supervised learning, a model learns from labeled data (e.g., social media messages labeled by sentiment) to infer (or predict) these labels from unlabeled texts. In contrast, unsupervised learning uses unlabeled data. Supervised approaches are especially suitable when we have a clear definition of the concept of interest and when labels are available (either annotated or native to the data). Unsupervised approaches, such as topic models, are especially useful for exploration. In this setting, conceptualization and operationalization may occur simultaneously, with theory emerging from the data BIBREF26 . Unsupervised approaches are also used when there is a clear way of measuring a concept, often based on strong assumptions. For example, BIBREF3 measure “surprise” in an analysis of Darwin's reading decisions based on the divergence between two probability distributions.
From an analysis perspective, the unit of text that we are labeling (or annotating, or coding), either automatic or manual, can sometimes be different than one's final unit of analysis. For example, if in a study on media frames in news stories, the theoretical framework and research question point toward frames at the story level (e.g., what is the overall causal analysis of the news article?), the story must be the unit of analysis. Yet it is often difficult to validly and reliably code a single frame at the story level. Multiple perspectives are likely to sit side-by-side in a story. Thus, an article on income inequality might point to multiple causes, such as globalization, education, and tax policies. Coding at the sentence level would detect each of these causal explanations individually, but this information would need to be somehow aggregated to determine the overall story-level frame. Sometimes scholars solve this problem by only examining headlines and lead paragraphs, arguing that based on journalistic convention, the most important information can be found at the beginning of a story. However, this leads to a return to a shorter, less nuanced analysis.
From a computational perspective, the unit of text can also make a huge difference, especially when we are using bag-of-words models, where word order within a unit does not matter. Small segments, like tweets, sometimes do not have enough information to make their semantic context clear. In contrast, larger segments, like novels, have too much variation, making it difficult to train focused models. Finding a good segmentation sometimes means combining short documents and subdividing long documents. The word “document" can therefore be misleading. But it is so ingrained in the common NLP lexicon that we use it anyway in this article.
For insight-driven text analysis, it is often critical that high-level patterns can be communicated. Furthermore, interpretable models make it easier to find spurious features, to do error analysis, and to support interpretation of results. Some approaches are effective for prediction, but harder to interpret. The value we place on interpretability can therefore influence the approach we choose. There is an increasing interest in developing interpretable or transparent models in the NLP and machine learning communities.
Annotation
Many studies involve human coders. Sometimes the goal is to fully code the data, but in a computational analysis we often use the labels (or annotations) to train machine learning models to automatically recognize them, and to identify language patterns that are associated with these labels. For example, for a project analyzing rumors online BIBREF27 , conversation threads were annotated along different dimensions, including rumor versus non-rumor and stance towards a rumor.
The collection of annotation choices make up an annotation scheme (or “codebook”). Existing schemes and annotations can be useful as starting points. Usually settling on an annotation scheme requires several iterations, in which the guidelines are updated and annotation examples are added. For example, a political scientist could use a mixed deductive-inductive strategy for developing a codebook. She starts by laying out a set of theory-driven deductive coding rules, which means that the broad principles of the coding rules are laid out without examining examples first. These are then tested (and possibly adjusted) based on a sample of the data. In line with Adcock and Collier's notion of “content validity” BIBREF13 , the goal is to assess whether the codebook adequately captures the systematized concept. By looking at the data themselves, she gains a better sense of whether some things have been left out of the coding rules and whether anything is superfluous, misleading, or confusing. Adjustments are made and the process is repeated, often with another researcher involved.
The final annotations can be collected using a crowdsourcing platform, a smaller number of highly-trained annotators, or a group of experts. Which type of annotator to use should be informed by the complexity and specificity of the concept. For more complex concepts, highly-trained or expert annotators tend to produce more reliable results. However, complex concepts can sometimes be broken down into micro-tasks that can be performed independently in parallel by crowdsourced annotators. Concepts from highly specialized domains may require expert annotators. In all cases, however, some training will be required, and the training phase should involve continual checks of inter-annotator agreement (i.e. intercoder reliability) or checks against a gold standard (e.g. quizzes in crowdsourcing platforms).
We also need to decide how inter-annotator agreement will be measured and what an acceptable level of agreement would be. Krippendorff's alpha is frequently used in the social sciences, but the right measure depends on the type of data and task. For manual coding, we can continually check inter-annotator agreement and begin introducing checks of intra-annotator agreement, too. For most communication scholars using only manual content analysis, an acceptable rate of agreement is achieved when Krippendorf's alpha reaches 0.80 or above. When human-coded data are used to validate machine learning algorithms, the reliability of the human-coded data is even more important. Disagreement between annotators can signal weaknesses of the annotation scheme, or highlight the inherent ambiguity in what we are trying to measure. Disagreement itself can be meaningful and can be integrated in subsequent analyses BIBREF28 , BIBREF29 .
Data pre-processing
Preparing the data can be a complex and time-consuming process, often involving working with partially or wholly unstructured data. The pre-processing steps have a big impact on the operationalizations, subsequent analyses and reproducibility efforts BIBREF30 , and they are usually tightly linked to what we intend to measure. Unfortunately, these steps tend to be underreported, but documenting the pre-processing choices made is essential and is analogous to recording the decisions taken during the production of a scholarly edition or protocols in biomedical research. Data may also vary enormously in quality, depending on how it has been generated. Many historians, for example, work with text produced from an analogue original using Optical Character Recognition (OCR). Often, there will be limited information available regarding the accuracy of the OCR, and the degree of accuracy may even vary within a single corpus (e.g. where digitized text has been produced over a period of years, and the software has gradually improved). The first step, then, is to try to correct for common OCR errors. These will vary depending on the type of text, the date at which the `original' was produced, and the nature of the font and typesetting.
One step that almost everyone takes is to tokenize the original character sequence into the words and word-like units. Tokenization is a more subtle and more powerful process than people expect. It is often done using regular expressions or scripts that have been circulating within the NLP community. Tokenization heuristics, however, can be badly confused by emoticons, creative orthography (e.g., U$A, sh!t), and missing whitespace. Multi-word terms are also challenging. Treating them as a single unit can dramatically alter the patterns in text. Many words that are individually ambiguous have clear, unmistakable meanings as terms, like “black hole" or “European Union". However, deciding what constitutes a multi-word term is a difficult problem. In writing systems like Chinese, tokenization is a research problem in its own right.
Beyond tokenization, common steps include lowercasing, removing punctuation, stemming (removing suffixes), lemmatization (converting inflections to a base lemma), and normalization, which has never been clearly defined, but often includes grouping abbreviations like “U.S.A." and “USA", ordinals like “1st" and “first", and variant spellings like “noooooo". The main goal of these steps is to improve the ratio of tokens (individual occurrences) to types (the distinct things in a corpus). Each step requires making additional assumptions about which distinctions are relevant: is “apple” different from “Apple”? Is “burnt” different from “burned”? Is “cool" different from “coooool"? Sometimes these steps can actively hide useful patterns, like social meaning BIBREF32 . Some of us therefore try do as little modification as possible.
From a multilingual perspective, English and Chinese have an unusually simple inflectional system, and so it is statistically reasonable to treat each inflection as a unique word type. Romance languages have considerably more inflections than English; many indigenous North American languages have still more. For these languages, unseen data is far more likely to include previously-unseen inflections, and therefore, dealing with inflections is more important. On the other hand, the resources for handling inflections vary greatly by language, with European languages dominating the attention of the computational linguistics community thus far.
We sometimes also remove words that are not relevant to our goals, for example by calculating vocabulary frequencies. We construct a “stoplist” of words that we are not interested in. If we are looking for semantic themes we might remove function words like determiners and prepositions. If we are looking for author-specific styles, we might remove all words except function words. Some words are generally meaningful but too frequent to be useful within a specific collection. We sometimes also remove very infrequent words. Their occurrences are too low for robust patterns and removing them helps reducing the vocabulary size.
The choice of processing steps can be guided by theory or knowledge about the domain as well as experimental investigation. When we have labels, predictive accuracy of a model is a way to assess the effect of the processing steps. In unsupervised settings, it is more challenging to understand the effects of different steps. Inferences drawn from unsupervised settings can be sensitive to pre-processing choices BIBREF33 . Stemming has been found to provide little measurable benefits for topic modeling and can sometimes even be harmful BIBREF34 . All in all, this again highlights the need to document these steps.
Finally, we can also mark up the data, e.g., by identifying entities (people, places, organizations, etc.) or parts of speech. Although many NLP tools are available for such tasks, they are often challenged by linguistic variation, such as orthographic variation in historical texts BIBREF35 and social media BIBREF32 . Moreover, the performance of NLP tools often drops when applying them outside the training domain, such as applying tools developed on newswire texts to texts written by younger authors BIBREF36 . Problems (e.g., disambiguation in named entity recognition) are sometimes resolved using considerable manual intervention. This combination of the automated and the manual, however, becomes more difficult as the scale of the data increases, and the `certainty' brought by the latter may have to be abandoned.
Dictionary-based approaches
Dictionaries are frequently used to code texts in content analyses BIBREF37 . Dictionaries consist of one or more categories (i.e. word lists). Sometimes the output is simply the number of category occurrences (e.g., positive sentiment), thus weighting words within a category equally. In some other cases, words are assigned continuous scores. The high transparency of dictionaries makes them sometimes more suitable than supervised machine learning models. However, dictionaries should only be used if the scores assigned to words match how the words are used in the data (see BIBREF38 for a detailed discussion on limitations). There are many off-the-shelf dictionaries available (e.g., LIWC BIBREF39 ). These are often well-validated, but applying them on a new domain may not be appropriate without additional validation. Corpus- or domain-specific dictionaries can overcome limitations of general-purpose dictionaries.
The dictionaries are often manually compiled, but increasingly they are constructed semi-automatically (e.g., BIBREF40 ). When we semi-automatically create a word list, we use automation to identify an initial word list, and human insight to filter it. By automatically generating the initial words lists, words can be identified that human annotators might have difficulty intuiting. By manually filtering the lists, we use our theoretical understanding of the target concept to remove spurious features.
In the introduction study, SAGE BIBREF41 was used to obtain a list of words that distinguished the text in the treatment group (subreddits that were closed by Reddit) from text in the control group (similar subreddits that were not closed). The researchers then returned to the hate speech definition provided by the European Court of Human Rights, and manually filtered the top SAGE words based on this definition. Not all identified words fitted the definition. The others included: the names of the subreddits themselves, names of related subreddits, community-specific jargon that was not directly related to hate speech, and terms such as IQ and welfare, which were frequently used in discourses of hate speech, but had significant other uses. The word lists provided the measurement instrument for their main result, which is that the use of hate speech throughout Reddit declined after the two treatment subreddits were closed.
Supervised models
Supervised learning is frequently used to scale up analyses. For example, BIBREF42 wanted to analyze the motivations of Movember campaign participants. By developing a classifier based on a small set of annotations, they were able to expand the analysis to over 90k participants.
The choice of supervised learning model is often guided by the task definition and the label types. For example, to identify stance towards rumors based on sequential annotations, an algorithm for learning from sequential BIBREF43 or time series data BIBREF44 could be used. The features (sometimes called variables or predictors) are used by the model to make the predictions. They may vary from content-based features such as single words, sequences of words, or information about their syntactic structure, to meta-information such as user or network information. Deciding on the features requires experimentation and expert insight and is often called feature engineering. For insight-driven analysis, we are often interested in why a prediction has been made and features that can be interpreted by humans may be preferred. Recent neural network approaches often use simple features as input (such as word embeddings or character sequences), which requires less feature engineering but make interpretation more difficult.
Supervised models are powerful, but they can latch on to spurious features of the dataset. This is particularly true for datasets that are not well-balanced, and for annotations that are noisy. In our introductory example on hate speech in Reddit BIBREF0 , the annotations are automatically derived from the forum in which each post appears, and indeed, many of the posts in the forums (subreddits) that were banned by Reddit would be perceived by many as hate speech. But even in banned subreddits, not all of the content is hate speech (e.g., some of the top features were self-referential like the name of the subreddit) but a classifier would learn a high weight for these features.
Even when expert annotations are available on the level of individual posts, spurious features may remain. BIBREF45 produced expert annotations of hate speech on Twitter. They found that one of the strongest features for sexism is the name of an Australian TV show, because people like to post sexist comments about the contestants. If we are trying to make claims about what inhibits or encourages hate speech, we would not want those claims to be tied to the TV show's popularity. Such problems are inevitable when datasets are not well-balanced over time, across genres, topics, etc. Especially with social media data, we lack a clear and objective definition of `balance' at this time.
The risk of supervised models latching on to spurious features reinforces the need for interpretability. Although the development of supervised models is usually performance driven, placing more emphasis on interpretability could increase the adoption of these models in insight-driven analyses. One way would be to only use models that are already somewhat interpretable, for example models that use a small number of human-interpretable features. Rather than imposing such restrictions, there is also work on generating post-hoc explanations for individual predictions (e.g., BIBREF46 ), even when the underlying model itself is very complex.
Topic modeling
Topic models (e.g., LDA BIBREF47 ) are usually unsupervised and therefore less biased towards human-defined categories. They are especially suited for insight-driven analysis, because they are constrained in ways that make their output interpretable. Although there is no guarantee that a “topic” will correspond to a recognizable theme or event or discourse, they often do so in ways that other methods do not. Their easy applicability without supervision and ready interpretability make topic models good for exploration. Topic models are less successful for many performance-driven applications. Raw word features are almost always better than topics for search and document classification. LSTMs and other neural network models are better as language models. Continuous word embeddings have more expressive power to represent fine-grained semantic similarities between words.
A topic model provides a different perspective on a collection. It creates a set of probability distributions over the vocabulary of the collection, which, when combined together in different proportions, best match the content of the collection. We can sort the words in each of these distributions in descending order by probability, take some arbitrary number of most-probable words, and get a sense of what (if anything) the topic is “about”. Each of the text segments also has its own distribution over the topics, and we can sort these segments by their probability within a given topic to get a sense of how that topic is used.
One of the most common questions about topic models is how many topics to use, usually with the implicit assumption that there is a “right” number that is inherent in the collection. We prefer to think of this parameter as more like the scale of a map or the magnification of a microscope. The “right” number is determined by the needs of the user, not by the collection. If the analyst is looking for a broad overview, a relatively small number of topics may be best. If the analyst is looking for fine-grained phenomena, a larger number is better.
After fitting the model, it may be necessary to circle back to an earlier phase. Topic models find consistent patterns. When authors repeatedly use a particular theme or discourse, that repetition creates a consistent pattern. But other factors can also create similar patterns, which look as good to the algorithm. We might notice a topic that has highest probability on French stopwords, indicating that we need to do a better job of filtering by language. We might notice a topic of word fragments, such as “ing”, “tion”, “inter”, indicating that we are not handling end-of-line hyphenation correctly. We may need to add to our stoplist or change how we curate multi-word terms.
Validation
The output of our measurement procedures (in the social sciences often called the “scores”) must now be assessed in terms of their reliability and validity with regard to the (systemized) concept. Reliability aims to capture repeatability, i.e. the extent to which a given tool provides consistent results.
Validity assesses the extent to which a given measurement tool measures what it is supposed to measure. In NLP and machine learning, most models are primarily evaluated by comparing the machine-generated labels against an annotated sample. This approach presumes that the human output is the “gold standard" against which performance should be tested. In contrast, when the reliability is measured based on the output of different annotators, no coder is taken as the standard and the likelihood of coders reaching agreement by chance (rather than because they are “correct") is factored into the resulting statistic. Comparing against a “gold standard” suggests that the threshold for human inter- and intra-coder reliability should be particularly high.
Accuracy, as well as other measures such as precision, recall and F-score, are sometimes presented as a measure of validity, but if we do not have a genuinely objective determination of what something is supposed measure—as is often the case in text analysis—then accuracy is perhaps a better indication of reliability than of validity. In that case, validity needs to be assessed based on other techniques like those we discuss later in this section. It is also worth asking what level of accuracy is sufficient for our analysis and to what extent there may be an upper bound, especially when the labels are native to the data or when the notion of a “gold standard” is not appropriate.
For some in the humanities, validation takes the form of close reading, not designed to confirm whether the model output is correct, but to present what BIBREF48 refers to as a form of “further discovery in two directions”. Model outputs tell us something about the texts, while a close reading of the texts alongside those outputs tells us something about the models that can be used for more effective model building. Applying this circular, iterative process to 450 18th-century novels written in three languages, Piper was able to uncover a new form of “conversional novel” that was not previously captured in “literary history's received critical categories” BIBREF48 .
Along similar lines, we can subject both the machine-generated output and the human annotations to another round of content validation. That is, take a stratified random sample, selecting observations from the full range of scores, and ask: Do these make sense in light of the systematized concept? If not, what seems to be missing? Or is something extraneous being captured? This is primarily a qualitative process that requires returning to theory and interrogating the systematized concept, indicators, and scores together. This type of validation is rarely done in NLP, but it is especially important when it is difficult to assess what drives a given machine learning model. If there is a mismatch between the scores and systematized concept at this stage, the codebook may need to be adjusted, human coders retrained, more training data prepared, algorithms adjusted, or in some instances, even a new analytical method adopted.
Other types of validation are also possible, such as comparing with other approaches that aim to capture the same concept, or comparing the output with external measures (e.g., public opinion polls, the occurrence of future events). We can also go beyond only evaluating the labels (or point estimates). BIBREF16 used human judgments to not only assess the positional estimates from a scaling method of latent political traits but also to assess uncertainty intervals. Using different types of validation can increase our confidence in the approach, especially when there is no clear notion of ground truth.
Besides focusing on rather abstract evaluation measures, we could also assess the models in task-based settings using human experts. Furthermore, for insight-driven analyses, it can be more useful to focus on improving explanatory power than making small improvements in predictive performance.
Analysis
In this phase, we use our models to explore or answer our research questions. For example, given a topic model we can look at the connection between topics and metadata elements. Tags such as “hate speech" or metadata information imply a certain way of organizing the collection. Computational models provide another organization, which may differ in ways that provide more insight into how these categories manifest themselves, or fail to do so.
Moreover, when using a supervised approach, the “errors”, i.e. disagreement between the system output and human-provided labels, can point towards interesting cases for closer analysis and help us reflect on our conceptualizations. In the words of BIBREF2 , they can be “opportunities for interpretation”. Other types of “failures” can be insightful as well. Sometimes there is a “dog that didn't bark” BIBREF49 –i.e., something that everyone thinks we should have found, but we did not. Or, sometimes the failures are telling us about the existence of something in the data that nobody noticed, or thought important, until then (e.g., the large number of travel journals in Darwin's reading lists).
Computational text analysis is not a replacement for but rather an addition to the approaches one can take to analyze social and cultural phenomena using textual data. By moving back and forth between large-scale computational analyses and small-scale qualitative analyses, we can combine their strengths so that we can identify large-scale and long-term trends, but also tell individual stories. For example, the Reddit study on hate speech BIBREF0 raised various follow-up questions: Can we distinguish hate speech from people talking about hate speech? Did people find new ways to express hate speech? If so, did the total amount of online hate speech decrease after all? As possible next steps, a qualitative discourse analyst might examine a smaller corpus to investigate whether commenters were indeed expressing hate speech in new ways; a specialist in interview methodologies might reach out to commenters to better understand the role of online hate speech in their lives. Computational text analysis represents a step towards better understanding social and cultural phenomena, and it is in many cases better suited towards opening questions rather than closing them.
Conclusion
Insight-driven computational analysis of text is becoming increasingly common. It not only helps us see more broadly, it helps us see subtle patterns more clearly and allows us to explore radical new questions about culture and society. In this article we have consolidated our experiences, as scholars from very different disciplines, in analyzing text as social and cultural data and described how the research process often unfolds. Each of the steps in the process is time-consuming and labor-intensive. Each presents challenges. And especially when working across disciplines, the research often involves a fair amount of discussion—even negotiation—about what means of operationalization and approaches to analysis are appropriate and feasible. And yet, with a bit of perseverance and mutual understanding, conceptually sound and meaningful work results so that we can truly make use of the exciting opportunities rich textual data offers.
Acknowledgements
This work was supported by The Alan Turing Institute under the EPSRC grant EP/N510129/1. Dong Nguyen is supported with an Alan Turing Institute Fellowship (TU/A/000006). Maria Liakata is a Turing fellow at 40%. We would also like to thank the participants of the “Bridging disciplines in analysing text as social and cultural data” workshop held at the Turing Institute (2017) for insightful discussions. The workshop was funded by a Turing Institute seed funding award to Nguyen and Liakata.
Question:
What kind of issues (that are not on the forefront of computational text analysis) do they tackle?
Answer:
| identifying the questions we wish to explore, Can text analysis provide a new perspective on a “big question” that has been attracting interest for years?, How can we explain what we observe?, hope to connect to multiple disciplines | 328 | 55,834 | 55,836 | 55,944 | ... [The rest of the article is omitted]
|
c70bafc35e27be9d1efae60596bc0dd390c124c0 | c70bafc35e27be9d1efae60596bc0dd390c124c0_0 | You are given a scientific article and a question. Answer the question as concisely as you can, using a single phrase or sentence if possible. If the question cannot be answered based on the information in the article, write "unanswerable". If the question is a yes/no question, answer "yes", "no", or "unanswerable".
Article:
Introduction
Privacy policies are the documents which disclose the ways in which a company gathers, uses, shares and manages a user's data. As legal documents, they function using the principle of notice and choice BIBREF0, where companies post their policies, and theoretically, users read the policies and decide to use a company's products or services only if they find the conditions outlined in its privacy policy acceptable. Many legal jurisdictions around the world accept this framework, including the United States and the European Union BIBREF1, BIBREF2. However, the legitimacy of this framework depends upon users actually reading and understanding privacy policies to determine whether company practices are acceptable to them BIBREF3. In practice this is seldom the case BIBREF4, BIBREF5, BIBREF6, BIBREF7, BIBREF8, BIBREF9, BIBREF10. This is further complicated by the highly individual and nuanced compromises that users are willing to make with their data BIBREF11, discouraging a `one-size-fits-all' approach to notice of data practices in privacy documents.
With devices constantly monitoring our environment, including our personal space and our bodies, lack of awareness of how our data is being used easily leads to problematic situations where users are outraged by information misuse, but companies insist that users have consented. The discovery of increasingly egregious uses of data by companies, such as the scandals involving Facebook and Cambridge Analytica BIBREF12, have further brought public attention to the privacy concerns of the internet and ubiquitous computing. This makes privacy a well-motivated application domain for NLP researchers, where advances in enabling users to quickly identify the privacy issues most salient to them can potentially have large real-world impact.
[1]https://play.google.com/store/apps/details?id=com.gotokeep.keep.intl [2]https://play.google.com/store/apps/details?id=com.viber.voip [3]A question might not have any supporting evidence for an answer within the privacy policy.
Motivated by this need, we contribute PrivacyQA, a corpus consisting of 1750 questions about the contents of privacy policies, paired with over 3500 expert annotations. The goal of this effort is to kickstart the development of question-answering methods for this domain, to address the (unrealistic) expectation that a large population should be reading many policies per day. In doing so, we identify several understudied challenges to our ability to answer these questions, with broad implications for systems seeking to serve users' information-seeking intent. By releasing this resource, we hope to provide an impetus to develop systems capable of language understanding in this increasingly important domain.
Related Work
Prior work has aimed to make privacy policies easier to understand. Prescriptive approaches towards communicating privacy information BIBREF21, BIBREF22, BIBREF23 have not been widely adopted by industry. Recently, there have been significant research effort devoted to understanding privacy policies by leveraging NLP techniques BIBREF24, BIBREF25, BIBREF26, BIBREF27, BIBREF28, especially by identifying specific data practices within a privacy policy. We adopt a personalized approach to understanding privacy policies, that allows users to query a document and selectively explore content salient to them. Most similar is the PolisisQA corpus BIBREF29, which examines questions users ask corporations on Twitter. Our approach differs in several ways: 1) The PrivacyQA dataset is larger, containing 10x as many questions and answers. 2) Answers are formulated by domain experts with legal training. 3) PrivacyQA includes diverse question types, including unanswerable and subjective questions.
Our work is also related to reading comprehension in the open domain, which is frequently based upon Wikipedia passages BIBREF16, BIBREF17, BIBREF15, BIBREF30 and news articles BIBREF20, BIBREF31, BIBREF32. Table.TABREF4 presents the desirable attributes our dataset shares with past approaches. This work is also tied into research in applying NLP approaches to legal documents BIBREF33, BIBREF34, BIBREF35, BIBREF36, BIBREF37, BIBREF38, BIBREF39. While privacy policies have legal implications, their intended audience consists of the general public rather than individuals with legal expertise. This arrangement is problematic because the entities that write privacy policies often have different goals than the audience. feng2015applying, tan-EtAl:2016:P16-1 examine question answering in the insurance domain, another specialized domain similar to privacy, where the intended audience is the general public.
Data Collection
We describe the data collection methodology used to construct PrivacyQA. With the goal of achieving broad coverage across application types, we collect privacy policies from 35 mobile applications representing a number of different categories in the Google Play Store. One of our goals is to include both policies from well-known applications, which are likely to have carefully-constructed privacy policies, and lesser-known applications with smaller install bases, whose policies might be considerably less sophisticated. Thus, setting 5 million installs as a threshold, we ensure each category includes applications with installs on both sides of this threshold. All policies included in the corpus are in English, and were collected before April 1, 2018, predating many companies' GDPR-focused BIBREF41 updates. We leave it to future studies BIBREF42 to look at the impact of the GDPR (e.g., to what extent GDPR requirements contribute to making it possible to provide users with more informative answers, and to what extent their disclosures continue to omit issues that matter to users).
Data Collection ::: Crowdsourced Question Elicitation
The intended audience for privacy policies consists of the general public. This informs the decision to elicit questions from crowdworkers on the contents of privacy policies. We choose not to show the contents of privacy policies to crowdworkers, a procedure motivated by a desire to avoid inadvertent biases BIBREF43, BIBREF44, BIBREF45, BIBREF46, BIBREF47, and encourage crowdworkers to ask a variety of questions beyond only asking questions based on practices described in the document.
Instead, crowdworkers are presented with public information about a mobile application available on the Google Play Store including its name, description and navigable screenshots. Figure FIGREF9 shows an example of our user interface. Crowdworkers are asked to imagine they have access to a trusted third-party privacy assistant, to whom they can ask any privacy question about a given mobile application. We use the Amazon Mechanical Turk platform and recruit crowdworkers who have been conferred “master” status and are located within the United States of America. Turkers are asked to provide five questions per mobile application, and are paid $2 per assignment, taking ~eight minutes to complete the task.
Data Collection ::: Answer Selection
To identify legally sound answers, we recruit seven experts with legal training to construct answers to Turker questions. Experts identify relevant evidence within the privacy policy, as well as provide meta-annotation on the question's relevance, subjectivity, OPP-115 category BIBREF49, and how likely any privacy policy is to contain the answer to the question asked.
Data Collection ::: Analysis
Table.TABREF17 presents aggregate statistics of the PrivacyQA dataset. 1750 questions are posed to our imaginary privacy assistant over 35 mobile applications and their associated privacy documents. As an initial step, we formulate the problem of answering user questions as an extractive sentence selection task, ignoring for now background knowledge, statistical data and legal expertise that could otherwise be brought to bear. The dataset is partitioned into a training set featuring 27 mobile applications and 1350 questions, and a test set consisting of 400 questions over 8 policy documents. This ensures that documents in training and test splits are mutually exclusive. Every question is answered by at least one expert. In addition, in order to estimate annotation reliability and provide for better evaluation, every question in the test set is answered by at least two additional experts.
Table TABREF14 describes the distribution over first words of questions posed by crowdworkers. We also observe low redundancy in the questions posed by crowdworkers over each policy, with each policy receiving ~49.94 unique questions despite crowdworkers independently posing questions. Questions are on average 8.4 words long. As declining to answer a question can be a legally sound response but is seldom practically useful, answers to questions where a minority of experts abstain to answer are filtered from the dataset. Privacy policies are ~3000 words long on average. The answers to the question asked by the users typically have ~100 words of evidence in the privacy policy document.
Data Collection ::: Analysis ::: Categories of Questions
Questions are organized under nine categories from the OPP-115 Corpus annotation scheme BIBREF49:
First Party Collection/Use: What, why and how information is collected by the service provider
Third Party Sharing/Collection: What, why and how information shared with or collected by third parties
Data Security: Protection measures for user information
Data Retention: How long user information will be stored
User Choice/Control: Control options available to users
User Access, Edit and Deletion: If/how users can access, edit or delete information
Policy Change: Informing users if policy information has been changed
International and Specific Audiences: Practices pertaining to a specific group of users
Other: General text, contact information or practices not covered by other categories.
For each question, domain experts indicate one or more relevant OPP-115 categories. We mark a category as relevant to a question if it is identified as such by at least two annotators. If no such category exists, the category is marked as `Other' if atleast one annotator has identified the `Other' category to be relevant. If neither of these conditions is satisfied, we label the question as having no agreement. The distribution of questions in the corpus across OPP-115 categories is as shown in Table.TABREF16. First party and third party related questions are the largest categories, forming nearly 66.4% of all questions asked to the privacy assistant.
Data Collection ::: Analysis ::: Answer Validation
When do experts disagree? We would like to analyze the reasons for potential disagreement on the annotation task, to ensure disagreements arise due to valid differences in opinion rather than lack of adequate specification in annotation guidelines. It is important to note that the annotators are experts rather than crowdworkers. Accordingly, their judgements can be considered valid, legally-informed opinions even when their perspectives differ. For the sake of this question we randomly sample 100 instances in the test data and analyze them for likely reasons for disagreements. We consider a disagreement to have occurred when more than one expert does not agree with the majority consensus. By disagreement we mean there is no overlap between the text identified as relevant by one expert and another.
We find that the annotators agree on the answer for 74% of the questions, even if the supporting evidence they identify is not identical i.e full overlap. They disagree on the remaining 26%. Sources of apparent disagreement correspond to situations when different experts: have differing interpretations of question intent (11%) (for example, when a user asks 'who can contact me through the app', the questions admits multiple interpretations, including seeking information about the features of the app, asking about first party collection/use of data or asking about third party collection/use of data), identify different sources of evidence for questions that ask if a practice is performed or not (4%), have differing interpretations of policy content (3%), identify a partial answer to a question in the privacy policy (2%) (for example, when the user asks `who is allowed to use the app' a majority of our annotators decline to answer, but the remaining annotators highlight partial evidence in the privacy policy which states that children under the age of 13 are not allowed to use the app), and other legitimate sources of disagreement (6%) which include personal subjective views of the annotators (for example, when the user asks `is my DNA information used in any way other than what is specified', some experts consider the boilerplate text of the privacy policy which states that it abides to practices described in the policy document as sufficient evidence to answer this question, whereas others do not).
Experimental Setup
We evaluate the ability of machine learning methods to identify relevant evidence for questions in the privacy domain. We establish baselines for the subtask of deciding on the answerability (§SECREF33) of a question, as well as the overall task of identifying evidence for questions from policies (§SECREF37). We describe aspects of the question that can render it unanswerable within the privacy domain (§SECREF41).
Experimental Setup ::: Answerability Identification Baselines
We define answerability identification as a binary classification task, evaluating model ability to predict if a question can be answered, given a question in isolation. This can serve as a prior for downstream question-answering. We describe three baselines on the answerability task, and find they considerably improve performance over a majority-class baseline.
SVM: We define 3 sets of features to characterize each question. The first is a simple bag-of-words set of features over the question (SVM-BOW), the second is bag-of-words features of the question as well as length of the question in words (SVM-BOW + LEN), and lastly we extract bag-of-words features, length of the question in words as well as part-of-speech tags for the question (SVM-BOW + LEN + POS). This results in vectors of 200, 201 and 228 dimensions respectively, which are provided to an SVM with a linear kernel.
CNN: We utilize a CNN neural encoder for answerability prediction. We use GloVe word embeddings BIBREF50, and a filter size of 5 with 64 filters to encode questions.
BERT: BERT BIBREF51 is a bidirectional transformer-based language-model BIBREF52. We fine-tune BERT-base on our binary answerability identification task with a learning rate of 2e-5 for 3 epochs, with a maximum sequence length of 128.
Experimental Setup ::: Privacy Question Answering
Our goal is to identify evidence within a privacy policy for questions asked by a user. This is framed as an answer sentence selection task, where models identify a set of evidence sentences from all candidate sentences in each policy.
Experimental Setup ::: Privacy Question Answering ::: Evaluation Metric
Our evaluation metric for answer-sentence selection is sentence-level F1, implemented similar to BIBREF30, BIBREF16. Precision and recall are implemented by measuring the overlap between predicted sentences and sets of gold-reference sentences. We report the average of the maximum F1 from each n$-$1 subset, in relation to the heldout reference.
Experimental Setup ::: Privacy Question Answering ::: Baselines
We describe baselines on this task, including a human performance baseline.
No-Answer Baseline (NA) : Most of the questions we receive are difficult to answer in a legally-sound way on the basis of information present in the privacy policy. We establish a simple baseline to quantify the effect of identifying every question as unanswerable.
Word Count Baseline : To quantify the effect of using simple lexical matching to answer the questions, we retrieve the top candidate policy sentences for each question using a word count baseline BIBREF53, which counts the number of question words that also appear in a sentence. We include the top 2, 3 and 5 candidates as baselines.
BERT: We implement two BERT-based baselines BIBREF51 for evidence identification. First, we train BERT on each query-policy sentence pair as a binary classification task to identify if the sentence is evidence for the question or not (Bert). We also experiment with a two-stage classifier, where we separately train the model on questions only to predict answerability. At inference time, if the answerable classifier predicts the question is answerable, the evidence identification classifier produces a set of candidate sentences (Bert + Unanswerable).
Human Performance: We pick each reference answer provided by an annotator, and compute the F1 with respect to the remaining references, as described in section 4.2.1. Each reference answer is treated as the prediction, and the remaining n-1 answers are treated as the gold reference. The average of the maximum F1 across all reference answers is computed as the human baseline.
Results and Discussion
The results of the answerability baselines are presented in Table TABREF31, and on answer sentence selection in Table TABREF32. We observe that bert exhibits the best performance on a binary answerability identification task. However, most baselines considerably exceed the performance of a majority-class baseline. This suggests considerable information in the question, indicating it's possible answerability within this domain.
Table.TABREF32 describes the performance of our baselines on the answer sentence selection task. The No-answer (NA) baseline performs at 28 F1, providing a lower bound on performance at this task. We observe that our best-performing baseline, Bert + Unanswerable achieves an F1 of 39.8. This suggest that bert is capable of making some progress towards answering questions in this difficult domain, while still leaving considerable headroom for improvement to reach human performance. Bert + Unanswerable performance suggests that incorporating information about answerability can help in this difficult domain. We examine this challenging phenomena of unanswerability further in Section .
Results and Discussion ::: Error Analysis
Disagreements are analyzed based on the OPP-115 categories of each question (Table.TABREF34). We compare our best performing BERT variant against the NA model and human performance. We observe significant room for improvement across all categories of questions but especially for first party, third party and data retention categories.
We analyze the performance of our strongest BERT variant, to identify classes of errors and directions for future improvement (Table.8). We observe that a majority of answerability mistakes made by the BERT model are questions which are in fact answerable, but are identified as unanswerable by BERT. We observe that BERT makes 124 such mistakes on the test set. We collect expert judgments on relevance, subjectivity , silence and information about how likely the question is to be answered from the privacy policy from our experts. We find that most of these mistakes are relevant questions. However many of them were identified as subjective by the annotators, and at least one annotator marked 19 of these questions as having no answer within the privacy policy. However, only 6 of these questions were unexpected or do not usually have an answer in privacy policies. These findings suggest that a more nuanced understanding of answerability might help improve model performance in his challenging domain.
Results and Discussion ::: What makes Questions Unanswerable?
We further ask legal experts to identify potential causes of unanswerability of questions. This analysis has considerable implications. While past work BIBREF17 has treated unanswerable questions as homogeneous, a question answering system might wish to have different treatments for different categories of `unanswerable' questions. The following factors were identified to play a role in unanswerability:
Incomprehensibility: If a question is incomprehensible to the extent that its meaning is not intelligible.
Relevance: Is this question in the scope of what could be answered by reading the privacy policy.
Ill-formedness: Is this question ambiguous or vague. An ambiguous statement will typically contain expressions that can refer to multiple potential explanations, whereas a vague statement carries a concept with an unclear or soft definition.
Silence: Other policies answer this type of question but this one does not.
Atypicality: The question is of a nature such that it is unlikely for any policy policy to have an answer to the question.
Our experts attempt to identify the different `unanswerable' factors for all 573 such questions in the corpus. 4.18% of the questions were identified as being incomprehensible (for example, `any difficulties to occupy the privacy assistant'). Amongst the comprehendable questions, 50% were identified as likely to have an answer within the privacy policy, 33.1% were identified as being privacy-related questions but not within the scope of a privacy policy (e.g., 'has Viber had any privacy breaches in the past?') and 16.9% of questions were identified as completely out-of-scope (e.g., `'will the app consume much space?'). In the questions identified as relevant, 32% were ill-formed questions that were phrased by the user in a manner considered vague or ambiguous. Of the questions that were both relevant as well as `well-formed', 95.7% of the questions were not answered by the policy in question but it was reasonable to expect that a privacy policy would contain an answer. The remaining 4.3% were described as reasonable questions, but of a nature generally not discussed in privacy policies. This suggests that the answerability of questions over privacy policies is a complex issue, and future systems should consider each of these factors when serving user's information seeking intent.
We examine a large-scale dataset of “natural” unanswerable questions BIBREF54 based on real user search engine queries to identify if similar unanswerability factors exist. It is important to note that these questions have previously been filtered, according to a criteria for bad questions defined as “(questions that are) ambiguous, incomprehensible, dependent on clear false presuppositions, opinion-seeking, or not clearly a request for factual information.” Annotators made the decision based on the content of the question without viewing the equivalent Wikipedia page. We randomly sample 100 questions from the development set which were identified as unanswerable, and find that 20% of the questions are not questions (e.g., “all I want for christmas is you mariah carey tour”). 12% of questions are unlikely to ever contain an answer on Wikipedia, corresponding closely to our atypicality category. 3% of questions are unlikely to have an answer anywhere (e.g., `what guides Santa home after he has delivered presents?'). 7% of questions are incomplete or open-ended (e.g., `the south west wind blows across nigeria between'). 3% of questions have an unresolvable coreference (e.g., `how do i get to Warsaw Missouri from here'). 4% of questions are vague, and a further 7% have unknown sources of error. 2% still contain false presuppositions (e.g., `what is the only fruit that does not have seeds?') and the remaining 42% do not have an answer within the document. This reinforces our belief that though they have been understudied in past work, any question answering system interacting with real users should expect to receive such unanticipated and unanswerable questions.
Conclusion
We present PrivacyQA, the first significant corpus of privacy policy questions and more than 3500 expert annotations of relevant answers. The goal of this work is to promote question-answering research in the specialized privacy domain, where it can have large real-world impact. Strong neural baselines on PrivacyQA achieve a performance of only 39.8 F1 on this corpus, indicating considerable room for future research. Further, we shed light on several important considerations that affect the answerability of questions. We hope this contribution leads to multidisciplinary efforts to precisely understand user intent and reconcile it with information in policy documents, from both the privacy and NLP communities.
Acknowledgements
This research was supported in part by grants from the National Science Foundation Secure and Trustworthy Computing program (CNS-1330596, CNS-1330214, CNS-15-13957, CNS-1801316, CNS-1914486, CNS-1914444) and a DARPA Brandeis grant on Personalized Privacy Assistants (FA8750-15-2-0277). The US Government is authorized to reproduce and distribute reprints for Governmental purposes not withstanding any copyright notation. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the NSF, DARPA, or the US Government. The authors would like to extend their gratitude to Elias Wright, Gian Mascioli, Kiara Pillay, Harrison Kay, Eliel Talo, Alexander Fagella and N. Cameron Russell for providing their valuable expertise and insight to this effort. The authors are also grateful to Eduard Hovy, Lorrie Cranor, Florian Schaub, Joel Reidenberg, Aditya Potukuchi and Igor Shalyminov for helpful discussions related to this work, and to the three anonymous reviewers of this draft for their constructive feedback. Finally, the authors would like to thank all crowdworkers who consented to participate in this study.
Question:
Are the answers double (and not triple) annotated?
Answer:
| Yes | 328 | 26,107 | 26,109 | 26,169 | ... [The rest of the article is omitted]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.