A Novel Developer Ranking Algorithm for Automatic Bug Triage Using Topic Model and Developer Relations
A Novel Developer Ranking Algorithm for Automatic Bug Triage Using Topic Model and Developer Relations
复制标题
DOI：
10.1109/apsec.2014.43
复制
发表时间：
2014-12
期刊：
影响因子：
--
通讯作者：
Zhang Tao;Geunseok Yang;Byungjeong Lee;E. Lua
中科院分区：
文献类型：
--
作者：
Zhang Tao;Geunseok Yang;Byungjeong Lee;E. Lua
最近，错误解决已成为软件维护的关键问题，其中推荐适当的修复程序是一项重要任务。存在一些可以实现自动错误分类（即开发人员推荐）的方法（例如社交网络和机器学习技术）。本文提出了一种新方法来推荐最合适的修复程序来解决错误。与以前的方法不同，所提出的方法结合了主题模型和开发人员关系（例如，Bug 报告者和受让人）来捕获开发人员对特定 bug 报告的兴趣和经验，我们可以在新的 bug 出现时安排最合适的开发人员来修复它。我们使用三个大型开源项目（包括 Eclipse、Mozilla Fire Fox 和 Net beans）来评估我们方法的性能。实验结果表明，我们的方法优于其他针对开发人员的推荐方法。
Recently, bug resolution has become a pivotal issue for software maintenance where recommendations for appropriate fixers are an important task. Some approaches (e.g., Social network and machine learning techniques) exist that can achieve automatic bug triage (i.e., Developer recommendation). This paper proposes a new method to recommend the most suitable fixer for bug resolution. Different from previous approaches, the proposed approaches combine topic model and developer relations (e.g., Bug reporter and assignee) to capture developers' interest and experience on specific bug reports, we can arrange for the most appropriate developer to fix a new bug when it comes in. We evaluate the performance of our method using three large-scale open-source projects, including Eclipse, Mozilla Fire fox, and Net beans. The experimental results reveal that our approach outperforms other recommendation methods for developers.