arXiv:2611.18581v1 Announce Type: new Abstract: Several studies have investigated the use of automated techniques to help practitioners identify security bug reports early in the defect management process. However, currently there is no comprehensive overview of this research area. Such an overview will help identify future research directions and consolidate the potentially relevant existing solutions. This study aims to provide a comprehensive overview of automated techniques for identifying security bug reports. To achieve this, we conducted a systematic mapping study and identified 28 primary studies. The study's findings show that the literature has evaluated various automatic techniques for identifying security bug reports, including traditional machine learning, deep learning, and more advanced large language models. The two main challenges identified are the lack of labeled data and data imbalance. Furthermore, we observe that existing studies lack evaluations in real industrial settings and rely mostly on archival data from open-source repositories. The most commonly used data sources in these studies include Wicket, Ambari, Camel, and Derby. Overall, the literature appears fragmented, highlighting the need for future work to focus on comprehensive evaluations of the most promising techniques.
Continue to arXiv Software Engineering →You are being forwarded in 6s — stay on this page