FixerCache: unsupervised caching active developers for diverse bug triage

Abstract

Context: Bug triage aims to recommend appropriate developers for new bugs in order to reduce time and effort in bug resolution. Most previous approaches for bug triage are supervised. Before recommending developers, these approaches need to learn developers' bug-fix preferences via building and training models using text-information of developers' historical bug reports. Goal: In this paper, we empirically address three limitations of supervised bug triage approaches and propose FixerCache, an unsupervised approach for bug triage by caching developers based on their activeness in components of products. Method: In FixerCache, each component of a product has a dynamic developer cache which contains prioritized developers according to developers' activeness scores. Given a new bug report, FixerCache recommends fixers with high activeness in developer cache to participate in fixing the new bug. Results: Results of experiments on four products from Eclipse and Mozilla show that FixerCache outperforms supervised bug triage approaches in both prediction accuracy and diversity. And it can achieve prediction accuracy up to 96.32% and diversity up to 91.67%, with top-10 recommendation list. Conclusions: FixerCache recommends fixers for new bugs based on developers' activeness in components of products with high prediction accuracy and diversity. Moreover, since FixerCache does not need to learn developers' bug-fix preferences through complex and time consuming processes, it could reduce bug triage time from hours of supervised approaches to seconds.