Datasets:
license: cc-by-4.0
task_categories:
- translation
language:
- zh
- en
tags:
- translation
- patent
- MT
size_categories:
- n<1K
🤝 Dataset Summary
The PatSnap Translation dataset contains Chinese-English parallel corpora from 268 full-text patents. All data is sourced from publicly available patent specifications. Each entry has been verified by human experts to ensure compliance with patent translation and drafting standards. This dataset constitutes part of the training data for Hiro-Translation.
🚀 Key Features
🌐 Bilingual Parallel Corpus — High-quality Chinese-English (ZH-EN) parallel sentence pairs designed for machine translation tasks.
📑 Patent Domain Specialization — Sourced from 268 full-text patents, capturing the specialized terminology, structure, and drafting conventions unique to patent documents.
✅ Human-Verified Quality — Every entry has been reviewed by human experts to ensure accuracy and compliance with professional patent translation and drafting standards.
🔓 Publicly Sourced & Open — All data originates from publicly available patent specifications, ensuring transparency and open accessibility.
🎯 Ideal Use Cases — Patent translation, technical/legal machine translation, terminology extraction, and domain adaptation research.
💻 Quick Start & MCP Integration
from datasets import load_dataset
dataset = load_dataset("PatSnap/Translation", split="train")
record = dataset[0]
print(record["cn"])
🔗 Resources
- Developer Portal: open.patsnap.com
- GitHub: github.com/patsnap