Spring Jira Bug Dataset

Bug Triage teams devote significant time to analyzing, evaluating, and approximating the priority andresolution time needed for reported bugs. These teams frequently suffer from issues such as a lack of domainand functional knowledge, leading to inaccuracies in priority assignments and impacting the resolutiontime of reported bugs. Therefore, autonomous solutions that can assist the triage team in prioritizing anddelivering bug-fix estimates are required. In this paper, various innovative multi-target models are proposedto predict bug priority and bug resolution time simultaneously. Several transformation methodologies areutilized, including classifier chains, binary relevance, label power sets, random k-label sets (Rakel), ensemblemethodology, and deep neural network models with functional API. The proposed multi-target models aretrained on bug summaries extracted from Jira Bug tracking software. The classification performance ofthese models is assessed using hamming loss, accuracy, weighted precision, weighted recall, and weightedF1-score measures. The results demonstrate significant improvements in prediction accuracy and efficiency,highlighting the potential of these models to enhance the bug triage process.