A Large Dataset of Video Game Patch Notes

Video games are complex and heterogeneous software systems that integrate diverse media artifacts and involve multidisciplinary development teams. Patch notes (developer-authored summaries of game updates) offer valuable insights into how games change over time and how gameplay is refined. In this research, we present a large dataset of patch notes extracted from the Steam platform, one of the largest digital distribution services for games. As Steam does not systematically separate patch notes from other news and announcements related to a game, we define a data extraction and filtering pipeline that processes over 145,000 games available via the Steam API. In total, we identified 1,085,168 patch notes from the 1,873,879 collected news items. The resulting dataset is enriched with game metadata to support future analyses by facilitating sampling and clustering. We document the methodology, provide the dataset and scripts for reproducibility, and discuss potential applications of the resources.电子游戏是一类复杂且异构的软件系统，集成了多元媒体素材，且需多学科开发团队协同开发。更新补丁说明（Patch notes，即开发者撰写的游戏更新内容摘要）能够提供宝贵的研究视角，帮助理解游戏随时间的迭代历程与游戏玩法的优化细节。本研究构建了一个源自Steam平台的大型更新补丁说明数据集，Steam平台是全球规模最大的游戏数字分发服务平台之一。由于Steam并未系统地将更新补丁说明与游戏相关的其他新闻及公告区分开来，我们设计了一套数据提取与过滤流程，用于处理通过Steam应用程序接口（Steam API）可获取的超过14.5万款游戏。最终，我们从采集到的1,873,879条新闻内容中，共识别出1,085,168条更新补丁说明。本数据集补充了完整的游戏元数据，可辅助采样与聚类操作，为后续相关研究分析提供支撑。本文详细阐述了该数据集的构建方法，公开了数据集与可复现研究所需的代码脚本，并探讨了该资源的潜在应用场景。