Scholarlink

GitHub has become one of the most popular platforms for open source version control and collaboration. In 2017, GitHub introduced the “Topics” feature, which allows repository owners to add descriptive topics to better characterize their repositories. However, incorrect topic assignment can adversely impact a repository’s visibility to potential contributors. Compared to previous datasets, which delete low-frequency topics or map them to the frequent featured topics, our dataset retains all valuable topics, taking full account of the diversity of topics. In our work, we have collected the top 50,000 starred repositories on GitHub so far, along with their text information such as descriptions and README. Finally we collected information from 28,386 repositories with a total of 162,038 topics, covering 22,710 distinct topics. This extensive dataset supports various research applications, such as topic recommendation and trend analysis in open-source projects. Our dataset is available freely at https://github.com/Zzzzzhuzhiwei/StellarTop.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

StellarTop : An Integrated Multi-topic Dataset on GitHub Repositories

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Zhiwei Zhu,
  • Wenrui Huang,
  • Wei Wang

摘要

GitHub has become one of the most popular platforms for open source version control and collaboration. In 2017, GitHub introduced the “Topics” feature, which allows repository owners to add descriptive topics to better characterize their repositories. However, incorrect topic assignment can adversely impact a repository’s visibility to potential contributors. Compared to previous datasets, which delete low-frequency topics or map them to the frequent featured topics, our dataset retains all valuable topics, taking full account of the diversity of topics. In our work, we have collected the top 50,000 starred repositories on GitHub so far, along with their text information such as descriptions and README. Finally we collected information from 28,386 repositories with a total of 162,038 topics, covering 22,710 distinct topics. This extensive dataset supports various research applications, such as topic recommendation and trend analysis in open-source projects. Our dataset is available freely at https://github.com/Zzzzzhuzhiwei/StellarTop.