一句话：通过挖掘约1.3万个GitHub项目，发现UML在开源中仍使用不足，但随PlantUML/Mermaid等文本格式出现复兴迹象。
本文研究 UML 在开源软件中的演变与现状。UML 虽常被视为软件设计与文档的标准，并在高校课程中广泛使用，但在开源开发者中采纳有限，且面临文档可用性、过时与细节不足等问题。作者挖掘并分析约 1.3 万个 GitHub 项目，设计基于文件扩展名与内容的策略和启发式来识别 UML 文件，对二十年演变进行定量分析，考察 UML 流行度、使用 UML 的项目特征，以及 UML 制品的作者、创建者与维护者。结果证实 UML 仍未被充分利用，但也发现复兴迹象，其与 PlantUML、Mermaid 等人类可读文本格式的流行相吻合。论文最后讨论这一复兴带来的新挑战及其对 UML 未来的影响。
Since its inception, UML, the Unified Modeling Language, has been touted as the way to go when it comes to designing and documenting software systems. While being an integral part of many university software engineering programs, UML has found little consideration among developers, especially in open source software. Reasons for this include that UML shares some shortcomings with other forms of documentation (e.g., limited availability, outdatedness, inadequate level of detail). We present a study to investigate the evolution and the current situation regarding the use of UML in open source projects. We mined and analyzed ~ 13k GitHub projects, developing strategies and heuristics to identify UML files through their extensions and contents, for a quantitative analysis of two decades of evolution of the usage of UML. We explored the popularity of UML, derived characteristics of projects leveraging UML, and analyzed the authors, creators and maintainers, of UML artifacts. Our study confirms that UML is indeed still under-utilized. At the same time we found evidence of a resurgence coinciding with the popularity of human-readable text-based formats, defined and used by tools like PlantUML and Mermaid. We discuss how identifying and addressing the new challenges implied by this resurgence could impact the future of UML.
DOI 原文 ·
@inproceedings{paperbot8757,
title = {UML is Back. Or is it? Investigating the Past, Present, and Future of UML in Open Source Software},
author = {Joseph Romeo and Marco Raglianti and Csaba Nagy and Michele Lanza},
booktitle = {ICSE},
year = {2025},
doi = {10.1109/icse55347.2025.00155}
}