Abstract

Bug assignment plays a critical role in the bug fixing process. However, bug assignment can be a burden for projects receiving a large number of bug reports. If a bug is assigned to a developer who lacks sufficient expertise to appropriately address it, the software project can be adversely impacted in terms of quality, developer hours, and aggregate cost. An automated strategy that provides a list of developers ranked by suitability based on their development history and the development history of the project can help teams more quickly and more accurately identify the appropriate developer for a bug report, potentially resulting in an increase in productivity. To automate the process of assigning bug reports to the appropriate developer, several studies have employed an approach that combines natural language processing and information retrieval techniques to extract two categories of features: one targeting developers who have fixed similar bugs before and one targeting developers who have worked on source files similar to the description of the bug. As developers document their changes through their commit messages it represents another rich resource for profiling their expertise, as the language used in commit messages typically more closely matches the language used in bug reports. In this study, we have replicated the approach presented in [32] that applies a learning-to-rank technique to rank appropriate developers for each bug report. Additionally, we have extended the study by proposing an additional set of features to better profile a developer through their commit logs and through the API project descriptions referenced in their code changes. Furthermore, we explore the appropriateness of a joint recommendation approach employing a learning-to-rank technique and an ordinal regression technique. To evaluate our model, we have considered more than 10,000 bug reports with their appropriate assignees. The experimental results demonstrate the efficiency of our model in comparison with the state-of-the-art methods in recommending developers for open bug reports.

Publication Date

4-2019

Document Type

Thesis

Student Type

Graduate

Degree Name

Software Engineering (MS)

Department, Program, or Center

Software Engineering (GCCIS)

Advisor

Mohamed Wiem Mkaouer

Advisor/Committee Member

Christian Newman

Advisor/Committee Member

Pradeep Murukannaiah

Campus

RIT – Main Campus

Share

COinS