An Empirical Study on Bug Assignment Automation Using Chinese Bug Data

Bug assignment is an important step in bug life-cycle management. In large projects, this task would consume a substantial amount of human effort. To compare with the previous studies on automatic bug assignment in FOSS (Free/Open Source Software) projects, we conduct a case study on a proprietary software project in China. Our study consists of two experiments of automatic bug assignment, using Chinese text and the other non-text information of bug data respectively. Based on text data of the bug repository, the first experiment uses SVM to predict bug assignments and achieve accuracy close to that by human triagers. The second one explores the usefulness of non-text data in making such prediction. The main results from our study includes that text data are most useful data in the bug tracking system to triage bugs, and automation based on text data could effectively reduce the manual effort.