Aliyah-20146588/github-issues

GitHub Issues Dataset - Hugging Face Datasets Repository Dataset Description This dataset contains 2,000 issues and pull requests from the Hugging Face Datasets GitHub repository. It includes comprehensive metadata about each issue/PR along with all associated comments, making it suitable for various NLP and software engineering tasks.

Dataset Summary Repository: huggingface/datasets Total Examples: 2,000 (1,000 train / 1,000 test) Collection Date: November 4, 2025 Language: English Data Source: GitHub REST API v3

Supported Tasks Issue Classification: Categorize issues by type, priority, or topic Automated Triage: Predict issue labels or assignment Comment Generation: Generate helpful responses to issues Summarization: Create concise summaries of lengthy discussions Sentiment Analysis: Analyze community sentiment and tone Duplicate Detection: Identify similar or duplicate issues Resolution Prediction: Predict if/when issues will be closed

Dataset Structure Data Instances Each example contains:

Issue/PR metadata (title, state, dates, author) User information Labels and assignees Complete comment thread Reaction counts Pull request linkage (if applicable)

Data Fields: Includes Field(Left), type(middle) and description (right)

url string API URL for the issue,

html_url string Web URL for the issue,

number int Issue/PR number,

title string Issue/PR title,

body string Issue/PR description,

state string Current state (open/closed),

created_at string Creation timestamp (ISO 8601),

updated_at string Last update timestamp,

closed_at string Closure timestamp (if closed),

user string Author information (JSON),

assignee string Assigned user (JSON),

assignees list[string] All assigned users,

labels list[string] Applied labels,

comments list[string] All comment bodies,

comment_count int Number of comments,

is_pull_request bool Whether this is a PR,

pull_request string PR metadata (JSON, if applicable),

reactions string Reaction counts (JSON),

locked bool Whether discussion is locked,

author_association string User's relationship to repo,

state_reason string Reason for current state

Data Splits Train: 1,000 examples (50%) Test: 1,000 examples (50%) Split performed randomly with seed for reproducibility.

Dataset Creation Source Data Data collected from the GitHub REST API (https://api.github.com) using authenticated requests.

Data Collection Process Fetched 2,000 most recent issues/PRs using pagination Filtered to include both open and closed items Retrieved all comments for each issue via separate API calls Added is_pull_request flag to distinguish PRs from issues Split dataset 50/50 into train and test sets

Annotations No manual annotations were added. All data comes directly from GitHub.

Licensing and Usage License This dataset contains data from GitHub and is subject to:

GitHub Terms of Service: https://docs.github.com/en/site-policy/github-terms/github-terms-of-service GitHub API Terms: https://docs.github.com/en/site-policy/github-terms/github-terms-for-additional-products-and-features#api-terms

Key License Considerations:

Public Data: All data was publicly accessible at time of collection API Usage: Collected via authenticated GitHub API following rate limits Redistribution: Subject to GitHub's terms regarding redistribution of public data Commercial Use: Review GitHub ToS for commercial usage restrictions

Privacy and Attribution User Information:

Contains usernames, avatars, and public profile links All information was publicly visible on GitHub No private or authenticated user data included Usernames should be treated as personally identifiable information (PII)

Attribution Requirements:

Original content created by GitHub users remains their intellectual property Dataset should be attributed to: "GitHub Issues from huggingface/datasets repository" Individual contributions belong to their respective authors

Recommended Citation:

Hugging Face Datasets GitHub Issues Dataset (2025) Source: https://github.com/huggingface/datasets Collected: November 4, 2025

Ethical Considerations Potential Biases Temporal Bias: Only reflects issues from a specific time period Community Bias: Reflects the Hugging Face Datasets user community Language Bias: Predominantly English language content Selection Bias: Most recent 2,000 issues may not represent all issue types

Privacy Considerations Public Information Only: All data was public at collection time User Privacy: Contains public usernames and profile information Content Sensitivity: May include bug reports with error messages or code snippets Right to be Forgotten: Users may delete/edit their comments after collection

Recommended Use Practices DO: Use for research and educational purposes Aggregate and anonymize when possible Respect user privacy in publications Check for updates to GitHub's terms

DON'T: Use for harassment or targeting individuals Assume data represents all software development Republish without proper attribution Use for training models that could harm users

Storage and Maintenance Storage Requirements Size: ~20.5 MB (10.2 MB train + 10.3 MB test) Format: JSONL (JSON Lines) Compression: None applied

Data Retention Snapshot Date: November 4, 2025 Update Frequency: Static snapshot (not automatically updated) Archival: Stored on Hugging Face Hub for long-term access

Known Limitations Static Snapshot: Issues/comments may have been edited or deleted since collection Incomplete Context: Does not include code changes, file attachments, or images API Limitations: Some fields may be truncated or sanitized by GitHub API Rate Limiting: Collection required rate limit management (5,000 requests/hour)

Technical Details Data Quality Completeness: All requested fields retrieved successfully Validation: Response codes verified (200 OK) Error Handling: Rate limit handling implemented with retry logic Duplicates: No duplicate issues (verified by issue number)

Preprocessing Minimal preprocessing applied JSON responses preserved in original structure Comments retrieved via separate API calls and added as list field Boolean flag added for pull request identification

Contact

For questions about this dataset:

Repository: Aliyah-20146588/github-issues Original Source: https://github.com/huggingface/datasets GitHub API Documentation: https://docs.github.com/en/rest

Changelog

Version 1.0 (November 4, 2025) Initial release 2,000 issues/PRs with comments Train/test split (1000/1000)