DRETOM: DEVELOPER RECOMMENDATION BASED ON TOPIC MODELS FOR BUG RESOLUTION

Background: In most cases, bug resolution is a collaborative activity among developers in software development where each developer contributes his or her ideas on how to resolve the bug. Although only one developer is recorded as the actual fixer for the bug, the contribution of the developers who participated in the collaboration cannot be neglected. Aims: This paper proposes a new approach, called DRETOM (Developer REcommendation based on TOpic Models), to recommending developers for bug resolution in collaborative behavior. Method: The proposed approach models developers’ interest in and expertise on bug resolving activities based on topic models that are built from their historical bug resolving records. Given a new bug report, DRETOM recommends a ranked list of developers who are potential to participate in and contribute to resolving the new bug according to these developers’ interest in and expertise on resolving it. Results: Experimental results on Eclipse JDT and Mozilla Firefox projects show that DRETOM can achieve high recall up to 82% and 50% with top 5 and top 7 recommendations respectively. Conclusion: Developers’ interest in bug resolving activities should be taken into consideration. On condition that the parameter θ of DRETOM is set properly with trials, the proposed approach is practically useful in terms of recall.