Dataset: Open Code, Hidden Collections: An Empirical Study of Academic Software Stewardship

This dataset accompanies the paper "Open Code, Hidden Collections: An Empirical Study of Academic Software Stewardship" (JCDL '26). It contains metadata for approximately 217,000 open source repositories affiliated with the 32 member institutions of CURIOSS (Community for University and Research Institution OSPOs), collected via the GitHub REST API and filtered using LLM-based classification. The dataset includes three files per university:- repositories.csv — per-repository metadata including institution, project type label (DEV, EDU, DATA, DOCS, WEB, OTHER), license, primary language, star/fork/subscriber counts, and presence of community files (README, Contributing Guide, Code of Conduct, Security Policy, Issue Templates, Pull Request Templates).- contributors.csv — contributor profiles for all the contributors of each repository.- organizations.csv — GitHub organization metadata for institutional organizations identified during discovery. Data was collected in June 2026. Pipeline code is available at https://github.com/juanis2112/JCDL-2026.