Ayushnangia/ana_v2_partial

UN Security Council Voting Data with Draft Resolutions

Overview

This dataset combines UN Security Council voting records with the full text of draft resolutions extracted from PDF documents. Each record represents a unique Security Council vote, including detailed member state votes, voting summaries, and the machine-readable text of the draft resolution document.

File: sc_voting_with_drafts.jsonl Format: JSON Lines (one JSON object per line) Size: 11 MB Records: 2,787 unique Security Council votes Time Period: 1946-2025 (79 years)

Data Sources

1. Voting Data: 2025_7_21_sc_voting.csv

   • 40,929 rows (15 member states × 2,787 votes)
   • Source: UN Digital Library voting records

2. Draft PDFs: pdf_draft/ directory

   • 488 draft resolution PDFs
   • All PDFs contain machine-readable text (99.6% extraction success rate)
   • 5.68 million characters extracted
   • Average: 11,640 characters per draft

Dataset Statistics

Coverage

• Total votes: 2,787
• With draft PDFs: 488 (17.5%)
• Without draft PDFs: 2,299 (82.5%)
• PDF extraction errors: 0

Temporal Distribution

Decade	Vote Records	Draft PDFs
1940s	78	0
1950s	54	0
1960s	143	1
1970s	186	5
1980s	185	1
1990s	638	73
2000s	623	157
2010s	596	180
2020s	284	71

Top Years by Draft Availability

1. 2002: 26 drafts
2. 1998: 25 drafts
3. 1999: 24 drafts
4. 2016: 22 drafts
5. 2006: 21 drafts

Data Structure

Each line is a JSON object with the following structure:

{
  "undl_id": "705442",
  "date": "2011-06-13",
  "resolution": "S/RES/1986(2011)",
  "draft": "S/2011/355",
  "meeting": "S/PV.6554",
  "description": "Security Council resolution 1986 (2011)...",
  "agenda": "The situation in Cyprus.",
  "subjects": "CYPRUS QUESTION",
  "vote_note": "",
  "modality": "Vote",
  "undl_link": "https://digitallibrary.un.org/record/705442",

  "vote_summary": {
    "total_yes": 15,
    "total_no": 0,
    "total_abstentions": 0,
    "total_non_voting": 0,
    "total_ms": 15
  },

  "member_state_votes": [
    {
      "ms_code": "CHN",
      "ms_name": "CHINA",
      "permanent_member": true,
      "vote": "Y"
    },
    ...
  ],

  "draft_pdf": {
    "has_pdf": true,
    "filename": "draft_S_2011_355.pdf",
    "text": "United Nations S/2011/355...",
    "char_count": 9740,
    "page_count": 4
  }
}

Field Descriptions

Core Metadata

• undl_id (string): Unique identifier from UN Digital Library
• date (string): Vote date in YYYY-MM-DD format
• resolution (string): Resolution number (e.g., "S/RES/1986(2011)")
• draft (string): Draft document number (e.g., "S/2011/355")
• meeting (string): Meeting record (e.g., "S/PV.6554")
• description (string): Full description of the resolution
• agenda (string): Agenda item description
• subjects (string): Subject classification
• vote_note (string): Additional voting notes
• modality (string): Voting type (typically "Vote")
• undl_link (string): URL to UN Digital Library record

Vote Summary

• total_yes (integer): Number of Yes votes
• total_no (integer): Number of No votes
• total_abstentions (integer): Number of abstentions
• total_non_voting (integer): Number of non-voting members
• total_ms (integer): Total member states (typically 15)

Member State Votes

Array of objects, one per Security Council member:

• ms_code (string): ISO country code (e.g., "USA", "CHN")
• ms_name (string): Country name (e.g., "UNITED STATES")
• permanent_member (boolean): Is this a permanent member (P5)?
• vote (string): Vote cast - "Y" (Yes), "N" (No), "A" (Abstain)

Draft PDF Data

• has_pdf (boolean): Whether draft PDF exists and was extracted
• filename (string|null): PDF filename (e.g., "draft_S_2011_355.pdf")
• text (string|null): Full extracted text from the PDF
• char_count (integer): Number of characters in extracted text
• page_count (integer): Number of pages in the PDF
• extraction_error (string|null): Error message if extraction failed

Usage Examples

Python

import json

# Load all records
records = []
with open('sc_voting_with_drafts.jsonl', 'r', encoding='utf-8') as f:
    for line in f:
        records.append(json.loads(line))

print(f"Loaded {len(records)} records")

# Find records with draft text
with_drafts = [r for r in records if r['draft_pdf']['has_pdf']]
print(f"Records with drafts: {len(with_drafts)}")

# Analyze voting patterns
unanimous = [r for r in records
             if r['vote_summary']['total_yes'] == r['vote_summary']['total_ms']]
print(f"Unanimous votes: {len(unanimous)}")

# Search draft text
keyword = "peacekeeping"
matching = [r for r in with_drafts
            if r['draft_pdf']['text'] and keyword.lower() in r['draft_pdf']['text'].lower()]
print(f"Drafts mentioning '{keyword}': {len(matching)}")

Command Line (jq)

# Count records by year
cat sc_voting_with_drafts.jsonl | jq -r '.date[:4]' | sort | uniq -c

# Find all vetoed resolutions
cat sc_voting_with_drafts.jsonl | jq 'select(.vote_summary.total_no > 0)'

# Extract all China votes
cat sc_voting_with_drafts.jsonl | jq '.member_state_votes[] | select(.ms_code == "CHN")'

# Get records with draft text over 50KB
cat sc_voting_with_drafts.jsonl | jq 'select(.draft_pdf.char_count > 50000)'

Analysis Ideas

1. Voting Pattern Analysis

   • P5 veto frequency and patterns
   • Voting bloc identification
   • Temporal trends in voting behavior

2. Text Analysis

   • Topic modeling on draft texts
   • Resolution complexity (by text length)
   • Language patterns in successful vs. failed resolutions

3. Correlation Studies

   • Resolution length vs. voting outcome
   • Subject matter vs. unanimity
   • Temporal trends in resolution topics

Data Quality Notes

PDF Text Extraction

• Success Rate: 100% (488/488 PDFs successfully extracted)
• Non-Readable PDFs: 2 identified in initial scan (not in this dataset)
• Extraction Method: PyPDF2 library
• Text Quality: Machine-readable, preserves line breaks and structure

Missing Data

• Draft PDFs: 82.5% of votes lack draft PDFs
  • Primarily older records (pre-1990s)
  • Some recent records also missing
• Empty Fields: Some records have empty agenda, subjects, or vote_note fields

Known Limitations

1. Older resolutions (1940s-1980s) have limited draft PDF availability
2. Some extracted text may contain formatting artifacts
3. Member state names use historical designations (e.g., "USSR", "ZAIRE")
4. Permanent member status reflects historical composition

File Generation

Script: create_voting_jsonl.py Generated: 2025-10-31 Processing Time: ~5 minutes for 2,787 records

Generation Process

1. Parse CSV voting data (40,929 rows)
2. Group by undl_id (2,787 unique votes)
3. For each vote:
   • Aggregate member state votes
   • Calculate vote summary
   • Extract draft PDF text if available
4. Write to JSONL format

License & Attribution

Data Source: United Nations Digital Library Website: https://digitallibrary.un.org/

Please cite the UN Digital Library when using this dataset. This compilation is provided for research and educational purposes.

Contact & Issues

For questions or issues with this dataset, please refer to the UN Digital Library documentation or the data processing scripts included in this repository.

---

Last Updated: October 31, 2025 Version: 1.0