An Optimization Technique for Unsupervised Automatic Extractive Bug Report Summarization

Abstract

Other profession individuals have a myth that the software development includes only programming element; in contrast, this line of work has strong element of system information management. The software organization requires the management and formulation of artifacts like designs, source code with documentation, specifications, and bug reports. These bug reports are stored into the software bug repositories. Software bug repositories have extreme information to perform the work on project. Tester’s needs to understand the artifacts related to project for recording the issues mentioned by customers and tracks the resolution of bugs. To find the main issue of a bug, a meaningful and huge conversation could happen between developer and reporter by bug reports. So the bug reports have a large amount of reported conversation in the form of messages from multiple peoples and these messages might contain few lines or multiple passages of unstructured text. For example, Mozilla bug #564,243 has 237 sentences and Mozilla bug #491,925 has 91 sentences as comments. To resolve the bug, the developer has to analyze all the comments in bug reports. This process can be tedious, time-consuming, burden, and very frustrating for the tester or developer. Sometimes, the amount of information might be overwhelming and sometimes, this information leads to duplicate, deserted, and non-optimized searches, just because the previous bug reports of the project has been ignored. In [1], the researcher suggested the way to reduce the developer effort and process time, consumed to identify the factual bug report is to provide bug summary.