Jon-Paul/github-issues1

GitHub Issues Dataset (huggingface/datasets)

Dataset Description

This dataset contains GitHub issues and pull requests from the huggingface/datasets repository. It was created using the GitHub REST API and processed into a machine-learning-friendly format using the Hugging Face datasets library.

Each record represents either:

• a GitHub issue, or
• a pull request (which GitHub treats as a special type of issue)

The dataset includes cleaned metadata fields and derived features to support NLP and machine learning tasks such as classification, summarisation, and exploratory analysis.

---

Dataset Summary

• Source: GitHub REST API (huggingface/datasets repository)
• Format: JSONL
• Total samples: ~8,000 (issues + pull requests combined)
• Language: English
• Created using: Python, Hugging Face datasets, GitHub API

---

Dataset Fields

Each entry contains the following fields:

• url: Direct GitHub URL to the issue or pull request
• number: Issue or PR number
• title: Title of the issue
• body: Full text content of the issue
• state: Issue state (open or closed)
• created_at: Creation timestamp
• comments_url: API URL for comments
• pull_request: If present, indicates the issue is a pull request (otherwise null)
• is_pull_request: Derived boolean flag (true if pull request, else false)
• text: Combined field of title + body for NLP tasks

---

Data Processing Pipeline

The dataset was built using the following steps:

1. Extracted issues using GitHub REST API
2. Saved raw responses into JSONL format
3. Cleaned nested and inconsistent fields
4. Removed problematic timestamp/nested structures when necessary
5. Created derived features:
   • is_pull_request
   • text (concatenated title and body)
6. Loaded using Hugging Face datasets.load_dataset

---

Intended Uses

This dataset is suitable for:

• Issue vs pull request classification
• NLP text classification tasks
• Summarisation of GitHub issues
• Repository analytics and insights
• Learning and experimenting with Hugging Face datasets

---

Limitations

• Contains data from only one repository (huggingface/datasets)
• Includes both issues and pull requests (must be filtered if not needed)
• Subject to GitHub API rate limits during data collection
• Text quality varies depending on user input in issues
• Not representative of all GitHub repositories

---

Ethical Considerations

• All data is publicly available on GitHub
• No private or sensitive user data is included beyond public usernames and contributions
• This dataset should not be used for profiling individual developers
• This project is done with the following in mind
• Australian Privacy Act 1988 (Cyber + Data Protection) https://www.oaic.gov.au/privacy/the-privacy-act
• ISO/IEC 27001 https://www.iso.org/isoiec-27001-information-security.html
• ISO/IEC 27002 – Security Controls Guidance https://www.iso.org/standard/75652.html

---

License

This dataset inherits the license of the original GitHub repository content (MIT License where applicable). Users should verify licensing constraints before commercial use.

---

Citation

If you use this dataset, please cite:

@dataset{github_issues_hf_datasets,
  author = {Jon-Paul Fitzgerald},
  title = {GitHub Issues Dataset (huggingface/datasets)},
  year = {2026},
  url = {https://github.com/huggingface/datasets}
}

@inproceedings{sanh2019distilbert,
  title={DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter},
  author={Sanh, Victor and Debut, Lysandre and Chaumond, Julien and Wolf, Thomas},
  year={2019},
  url ={https://huggingface.co/docs/transformers/main/en/model_doc/distilbert#transformers.DistilBertForSequenceClassification}
  eprint={1910.01108},
  archivePrefix={arXiv}
}