README.md

metadata

license: other
task_categories:
  - text-classification
  - sentence-similarity
  - feature-extraction
language:
  - en
tags:
  - github
  - issues
  - semantic-search
  - text-data
  - nlp
size_categories:
  - 10M<n<100M
pretty_name: GitHub Issues Dataset

Dataset Card for GitHub Issues Dataset

Dataset Description

This dataset contains raw GitHub issue records prepared for natural language processing tasks.

The dataset was created from GitHub issue data and includes text fields such as title, body, and comments. No major cleaning or filtering has been applied. In particular:

• Pull requests are NOT removed
• Missing or incomplete records are NOT removed
• The dataset reflects real-world, unclean data

A column named is_pull_request is included to indicate whether each record corresponds to a pull request or a standard issue.

The purpose of keeping the dataset in raw form is to allow preprocessing, cleaning, and feature engineering to be demonstrated as part of the analysis pipeline.

License

This dataset is derived from publicly available GitHub issue data.

The original content belongs to the respective repository owners and contributors.
This dataset is shared for educational and research purposes only.

Users must comply with GitHub Terms of Service and the licenses of the original repositories.

Dataset Structure

Each record represents one GitHub issue or pull request, including metadata, user information, and discussion content.

Data Fields

The dataset includes the following key fields:

Core Issue Information

• id: Unique identifier of the issue
• number: Issue number in the repository
• title: Title of the issue
• body: Main description text (may be empty)
• state: Status of the issue (open/closed)
• locked: Indicates if the issue is locked
• comments: List of comments associated with the issue

URLs and References

• url, html_url: API and web links to the issue
• repository_url: Link to the repository
• comments_url, events_url, labels_url: Related API endpoints

User Information

• user: Information about the issue creator (e.g., login, id)
• assignee: Assigned user (if available)
• assignees: List of assigned users
• closed_by: User who closed the issue

Labels and Tags

• labels: List of labels associated with the issue (name, color, description)

Time Information

• created_at: Issue creation timestamp
• updated_at: Last update timestamp
• closed_at: Issue closing timestamp

Pull Request Information

• pull_request: Contains pull request metadata if the record is a pull request
• is_pull_request: Boolean flag indicating whether the record is a pull request

Additional Metadata

• milestone: Milestone information (if available)
• author_association: Relationship of the author to the repository
• reactions: Reaction counts (likes, etc.)
• timeline_url: Timeline of issue events
• state_reason: Reason for closing (if available)

Data Preprocessing

This dataset is intentionally kept in raw format.

No major preprocessing steps such as removing missing values or filtering pull requests were applied.

However, a helper column (is_pull_request) was added to allow users to filter the dataset depending on their task.

Users of this dataset are expected to perform their own preprocessing steps, such as:

• Removing or handling missing values
• Filtering pull requests if needed
• Cleaning text data
• Combining fields (e.g., title + body + comments)

Intended Uses

This dataset is intended for educational and research purposes, particularly for:

• Natural Language Processing (NLP)
• Semantic search using embeddings (e.g., FAISS)
• Text similarity and clustering
• Feature extraction and data preprocessing exercises

The dataset is especially suitable for demonstrating data cleaning, feature engineering, and handling of raw, unstructured data.

Out-of-Scope Uses

This dataset should NOT be used for:

• Identifying or profiling individual GitHub users
• Making legal, financial, or employment decisions
• Training high-risk AI systems without proper data cleaning and validation
• Any use involving sensitive personal data without preprocessing and anonymisation

Limitations

• The dataset contains missing and incomplete text fields.
• Pull requests and issues are mixed together.
• Text data may include noise, informal language, or irrelevant content.
• Some records may not be useful without preprocessing.
• Public GitHub data may still contain unintended sensitive or personal information.

Because the dataset is raw, preprocessing is required before using it for machine learning tasks.

Ethical Considerations

Since this dataset is collected from public GitHub sources and kept in raw form:

• It may contain usernames or references to individuals
• It may include unintended sensitive information in issue text
• No automatic anonymisation has been applied

Users must review and clean the dataset before using it in any sensitive or production context.

Personal or sensitive information should be removed or masked where not necessary, in line with organisational AI and data governance policies.

Data Classification

This dataset is classified as Internal/Public (educational use).

Although the original data is publicly available, the processed dataset should still be handled responsibly. Data classification and handling should follow organisational rules for confidentiality, integrity, and access control