Bug Triaging is a vital part of issue managementsystems. Bug triaging deals with assigning a developer thetask of an incoming bug. This activity is error prone andtime consuming if done manually. There is a need forautomated support to accelerate this process. The currentautomated bug triaging systems exploits the text contents ofthe bug and the tossing relations among the developers. Theautomated bug triaging systems estimate the optimal bathbetween the first assignee of the bug and the bug resolverusing the tossing relations. The metrics used for assessingthe efficiency of bug triaging systems that are based ontossing relations is Mean number of Steps To Resolve(MSTR). This metric quantifies the number of steps reducedby the predicted path compared to the original path. It doesnot capture how far the retrieved path is in alignment withthe actual path. MSTR does reveal the informationregarding the extent to which the order of the developers inthe retrieved path is in line with that of the original path. Inaddition, there are no indicators for measuring the strengthof the retrieved path. In this paper, we propose two metrics(i) Path Similarity Metric which quantifies path alignmentbased on pair wise path alignment and (ii) Path AlignmentIndicator that measures the effectiveness of the retrievedpath based on degree centrality. The effectiveness of thetwo proposed metrics is validated using bug reportsextracted from the Eclipse project.
展开▼
