JustInternetAI/Graphitti-issues

Graphitti Issues and Pull Requests Dataset Summary

This dataset contains issues, pull requests, and their associated comments from the UW Bothell Graphitti Project, a high-speed graph simulator. It includes metadata such as:

Issue/PR metadata: title, body, state (open/closed/merged), labels, milestone, timestamps, author, etc.

Comments: discussion threads associated with each issue or PR.

Pull request-specific details: lines changed, commits, merged status, etc.

Use Cases:

Researchers can analyze open-source development practices, bug triaging, and collaboration dynamics.

Machine learning tasks like text classification (by label, severity, or topic), summarization (of long discussion threads), or sentiment analysis are all possible with these data.

Supported Tasks and Leaderboards

Potential tasks include:

Issue Classification / Triage
Classifying issues by type (bug, enhancement, question, etc.) or priority.

Topic Modeling
Discovering common themes or recurring issues in the project.

Summarization
Summarizing multi-turn issue or pull request discussions.

Sentiment Analysis
Analyzing the tone of contributors’ messages (helpful, frustrated, neutral, etc.).

Pull Request Analysis
Examining typical code contribution patterns or changes over time.

There are no official leaderboards associated with this dataset. Languages

Most text in this dataset is English, as GitHub issues and pull requests are typically written in English. Additionally, code snippets or references may appear in C++ (the Graphitti project’s primary language). Dataset Structure

Each row in this dataset corresponds to either an Issue or a Pull Request from the Graphitti repository. Key fields include:

id: The numerical GitHub ID.

type: Indicates issue or pull_request.

title: Title of the issue or PR.

body: Main description text.

state: For example, open, closed, merged.

user: GitHub user metadata (e.g., username, user ID).

comments: A sequence of comments (string text).

created_at, updated_at, closed_at: Timestamps in ISO format.

Additional fields like labels, milestone, reactions, etc.

Splits: According to the YAML front matter, the dataset is currently organized into a single train split containing 923 examples (issues/PRs). Users can choose to create validation/test splits as needed. Dataset Creation Curation Rationale

The Graphitti project is a significant academic and open-source effort at UW Bothell, focusing on high-speed graph simulation. Collecting issues and pull requests offers insight into:

Common bugs, feature requests, or usage patterns.

Developer collaboration and communication.

Code quality and testing practices in real-world open-source projects.

Source Data

Repository: UWB-Biocomputing/Graphitti on GitHub

Collection Method: The dataset was extracted via GitHub’s REST API (e.g., issues and pulls endpoints), capturing all available public issues and pull requests, along with their comment threads.

Annotations

No additional manual annotations: The labels, states, and statuses are those used in the original GitHub repository.

Who are the annotators?: All metadata and labels are generated by GitHub or by Graphitti’s contributors.

Considerations for Using the Data Social Impact

Can facilitate research on open-source community practices, academic software projects, and HPC (high-performance computing) collaboration patterns.

Offers an educational perspective on real-world software engineering processes.

Potential Risks and Ethical Considerations

Personal Data:

    GitHub usernames (and potentially user profiles, if included) could be considered personal data. While publicly available, re-publishing them in new contexts might raise privacy questions.

    Users may inadvertently share personal information within issue discussions.

Code Snippets:

    Issues or PRs may contain code under specific licenses (Graphitti is presumably under an open-source license). Ensure compliance with the original project’s license when reusing the code.

Toxic or Sensitive Language:

    While not likely in an academic project, some discussions could contain rude or offensive language. Screening is recommended if you plan to use it for moderation or sentiment tasks.

GitHub Terms of Service:

    This dataset is derived from content hosted on GitHub. Users should adhere to GitHub’s TOS regarding data usage and redistribution.

Discussion of Biases

The content reflects a specific project’s topics (high-speed graph simulation), so it may not generalize to other domains.

Language usage, patterns of communication, and culture are shaped by a small academic open-source community.

Additional Information Dataset Curators

UW Bothell CS Team (and community contributors). If you contributed to Graphitti’s issues or pull requests, you may be represented in the dataset. Licensing Information

Data License: GitHub repository data is generally subject to GitHub Terms of Service.

Code License: Graphitti’s source code is licensed under [ADD LICENSE HERE] (e.g., MIT, BSD, GPL?).

    If you redistribute code snippets, confirm it aligns with the project’s open-source license.

Citation Information

If you use this dataset, please cite:

 ```bibtex @misc{graphitti_issues_prs_2025, title = {Graphitti Issues and Pull Requests}, author = {UW Bothell CS Team}, howpublished = {Hugging Face, https://huggingface.co/datasets/YOUR_DATASET_SLUG}, year = {2025} } ``` 

Contributions

Special thanks to:

Justin Madison for dataset creation and curation.

Graphitti Maintainers and UW Bothell for hosting the original codebase and issue discussions.