Analyzing emotion words to predict severity of software bugs: a case study of open source projects

A successful software development project becomes an essential part of a software company's reputation. Thus, lots of project managers focus more on maintenance than on other management processes. Previous works studied how to help the maintenance process by detecting bug duplication and predicting the severity of bugs. This paper continues that kind of special work by analyzing emotion words for bug-severity prediction. In detail, we construct an emotion words-based dictionary for verifying bug reports' textual emotion analyses based on positive and negative terms. Then, we modify a machine learning algorithm, the Naïve Bayes multinomial, calling the new algorithm EWD-Multinomial. We compare this EWD-Multinomial study with our baselines, including Naïve Bayes multinomial and a Lamkanfi study, for open source projects such as Eclipse, Android, and JBoss. The result shows this study's algorithm outperforms the others.