Clustering for Bug Triage Based on Developer Work Capabilities

Triage of bugs is a crucial step in keeping software secure and encouraging prompt bug fixes. The majority of earlier research merely assigned the developers’ names to the classification label values produced by Bug Triage, which resulted in low accuracy, resource waste, and severe implications from reduced bug repair efficiency. This paper offers a novel solution to the aforementioned issues by grouping developers according to their work capabilities and using the clustering results as the bug triage results. This way, the developers can process the identified bugs and prevent resource waste and reduced bug repair efficiency because some developers are currently unavailable. This boosts bug repair efficiency significantly. We used the JENKINS dataset from the Eclipse open-source project for clustering tests and bug classification efforts. Compared to the traditional software bug assignment approach that does not consider developer clustering, the top-k accuracy has improved by 51.3%. The experimental findings demonstrate the reasonableness and efficacy of our clustering approach, which also significantly increases bug triage accuracy.