Reviewer Recommendation for Peer Review Based on File Path Similarity: A Case Study of the Android Open Source Project

Authors: Patanamon Thongtanunam, Ana E. C. Cruz, Norihiro Yoshida, Hajimu Iida

Venue: Knowledge-based Software Engineering (SIG – KBSE)

Abstract:

Peer Review is regarded as an important quality assurance mechanism in Open Source Software (OSS) development. Every change and its impact need to be assessed by developers to assure the quality of software. For effective code review, experience and skilled reviewers are required. However, in OSS peer review process, there is no explicit policy for selecting appropriate reviewers among available developers. The manual selection of those reviewers can be a costly and time-consuming task. To ease developers in OSS peer review process, we propose a reviewer recommendation algorithm determining similarity of file paths. As a case study, we recommended reviewers for the Android Open Source Project (AOSP). For this case study, our proposed algorithm obtained an accuracy of 75% while a previous approach obtained an accuracy of 30%.

Preprint: PDF