Towards more accurate severity prediction and fixer recommendation of software bugs

摘要: We propose REPtopic to search the top-K nearest neighbours of new bug report.New algorithms are developed predict severity and recommend fixers.Our approach performs better than previous works on two resolution tasks.REPtopic presents performance REP cosine similarity measures. Due unavoidable bugs appearing in most software systems, has become one important activities maintenance. For large-scale programs, developers usually depend reports fix given bugs. When a is reported, triager complete tasks that include identification fixer assignment. The purpose decide how quickly report should be addressed while assignment means needs assigned an appropriate developer for fixing. However, large number submitted every day increase triagers' workload, thus leading reduction accuracy Therefore it necessary develop automatic perform prediction recommendation instead manual work. This article proposes more accurate accomplish goal. firstly utilize modified algorithm (i.e., REPtopic) K-Nearest Neighbor (KNN) classification historical similar bug. Next, we extract their features (e.g., assignees similarity) algorithms. Finally, by adopting proposed algorithms, achieve semi-automatic five popular open source projects, including GNU Compiler Collection (GCC), OpenOffice, Eclipse, NetBeans, Mozilla. results demonstrated our method can improve through comparison with cutting-edge studies.