Software bug priority prediction technique based on intuitionistic fuzzy representation and class imbalance learning

In modern times, the software industry is more focused on the timely release of high-quality software. Software bugs have a significant impact on software quality and reliability. To complete the bug triaging process on time, the triager has to understand each bug and assign the correct priority to it. However, the bugs are reported rapidly, with lots of uncertainty and irregularities in the bug tracking system. Furthermore, there are multiple priority labels that are semantically close to each other. As a result, the triager is confused while understanding and prioritizing the bugs. To address these problems, the research presents an intuitionistic fuzzy representation of topic features-based software bug priority prediction (IFTBPP) technique. Initially, the imbalanced priority classes of software bugs are balanced using the synthetic minority oversampling technique. Then, topic modeling is used to create topics and terms for software bugs. The intuitionistic fuzzy set is used on the topics to compute various grades of a bug belonging to multiple priority classes. Finally, the similarity of a newly reported bug is calculated using intuitionistic fuzzy similarity measures with multiple priority classes. All the experiments of IFTBPP are conducted on Eclipse, Mozilla, Apache, and NetBeans repositories and compared with other existing models. The accuracy values obtained by IFTBPP on these repositories are 92.5%, 91.9%, 89.2%, and 93.9%, whereas the corresponding F-measure values are 91.7%, 91.3%, 88.9%, and 93.1%.