OSSGameBench: A Large-Scale Dataset of Contributor Activities in the Open-Source Video Games

Video games are a distinct type of software that engage users through deeply immersive and interactive experiences. Yet, unlike other types of software where standardized benchmark datasets have accelerated research on downstream tasks such as defect localization and repair, the game development domain lacks comparable resources that reflect its distinctive characteristics. For instance, defects in games often arise from the complex interplay between source code (e.g., C++ files) and non-code assets (e.g., graphics files), whereas in other software, such complications are less prevalent. Therefore, specialized datasets are required to capture these multidisciplinary demands unique to game development. Unavailability of such specialized datasets limits the analysis, development, and evaluation of data-driven solutions tailored to game development. To fill this void, we introduce OSSGameBench, a curated benchmark dataset derived from 950 open-source repositories in GitHub, encompassing both playable video games and essential game development tools (e.g., engines, frameworks). The dataset contains issues, comments, commits, pull requests, and review comments, with explicit mappings among these entities to enable future research aimed at improving the full spectrum of game development.

In this Zenodo repository, we will include the OSSGameBench dataset and the replication package, which contains the scripts used to collect and analyze the data. Please check the README.md file and our Online Appendix.pdf for detailed instructions on how to use our dataset and/or execute these scripts.

Please cite our paper:

@inproceedings{marsad2026msrdata,
  Author = {Marsad, Faiz and Weeraddana, Nimmi},
  Title = {{OSSGameBench: A Large-Scale Dataset of Contributor Activities in the Open-Source Video Games}},
  Year = {2026},
  Booktitle = {Proc. of the International Conference on Mining Software Repositories (MSR)}
}