A Data-aware Automatic Framework Of Bug Report Assignment

Bug report management is one of the important tasks of software development and maintenance process. Automatic bug report assignment, which is the important part of bug report management, can improve the efficiency of bug processing. This topic has been widely studied in the past decade, in which a usual method is to use text classification to accomplish assignment. Most of the existing efforts focus on the analysis on open-source bug report repositories, such as Bugzilla warehouse. The traditional heavy process of Bugzilla is difficult to satisfy the light development process and the rapid iterative requirement in the mobile Internet era, thus some lightweight Bug report management modes and systemes emerge as the times require.In order to adapt to the new Bug reports which are short and colloquial in the mobile Internet era, this paper presents a data-aware automatic assignment framework of Bug reports, namely DBug. DBug has three main steps for bug report assignment:data preprocessing, digital representation and automatic classification. In data preprocessing, the feature selection algorithm vF and feature selection proportion vFR are need to be set. In digital representation, the binding ratio of vector space model and topic statistical model vMR is need to be set. In automatic classification, the classification algorithm vC is need to be set. These four variables of DBug can be combined to achieve better assignment for different features of bug reports.In order to obtain the best practical values of four variables of DBug, (vF, vFR, vMR, vC), for different data sets of bug reports, we design and implement an empirical study on three types of data sets:Eclipse’s Bugzilla reports, crowdsourced test reports, Wechat reports of Mooctest. The preliminary experimental results show that (1) the variable setting can affect 10%-20% improvement of the accuracy of bug report assignement; (2) DBug works better crowdsourced test reports and Wechat reports that Bugzilla reports. DBug can be easily integrated into the Wechat public account, to fulfill a rapid intelligent bug report management system, so-called Mubug. We have implemented DBug in the mooctest project to verify its effectiveness.