Bug Report Summarization using Believability Score and Text Ranking

Abstract

During the maintenance phase of software development, bug reports provide software developers with important information. However, bug reports often include complex and long discussions. Therefore, concise and accurate summaries can help developers save the time for reading the full contents of bug reports. Several researchers have proposed summarizing bug reports. However, none of them proposed combining two different scores for measuring how important each sentence is among the developers' comments. In this paper, we propose an unsupervised bug report summarization which combines believability score and text ranking score for measuring the degree to which a sentence is important, in order to generate high-quality summaries. The experimental results over a public dataset show that our method outperforms the state-of-The-Art method in terms of summary quality.