Ranking of Bug Reports into Findbug Categories using the BM25 Function

Abstract

Bug reporting is a mechanism that is used by software companies to audit and track product quality. The developers and the users can log bug reports via a dedicated system. The quality of the bug reports affects the developers' endeavors to fix and close the bug. The reports include important information for the developers such as severity and priority. However, previous research found that most bugs are left without changing the severity level. Therefore, the developers need to know more information about the reported bug. In this research, we propose to improve the bug report by adding linking bug reports with a pre-specified bug description that was prepared by experts in bug types. i.e. the Findbug tool. The Findbug adds more information to the bug such as the bug type and bug category. The bugs are reclassified using a similarity function, BM25, to match it with more necessary information about the type of reported bugs. Text preprocessing was applied to remove unnecessary terms and to improve the search query in BM25. The proposed methodology could assign extra information to bug reports that help categorize them into different types such as correctness, security, performance, and others.