School of Software Technology, Dalian University of Technology, China

---

Abstract— Open source software USES software defect tracking system, which can effectively manage the related information of software defects, and build software defect data warehouse in the form of defect report.The severityattribute of software defect report can determine the important indicators such as the repairers, solving time and repairing rate of software defect.Much research on software defects focuses on severity analysis.In order to evaluate the workin the field of severity analysis, this paper reviews the existing studies.In particular, this paper introduces the main methods of severity study, and expounds the statistical characteristics analysis of severity attribute in software defect report dataset.According to the research status, the research of severity analysis is divided into qualitative analysis and quantitative analysis, and analyzed in detail.At the same time, each part is empirically analyzed based on data from the Mozilla project andEclipse project datasets.On this basis, this paper summarizes the existing work of severity analysis of defect report, and points out some possible problems in the work.

Index Terms— Software defect report severity,qualitative analysis,quantitative analysis, feature selection, machine learning.

[PDF]

Cite: Wenjie Liu, "Software Defect Data Mining: A Survey of Severity Analysis," Journal of Software vol. 14, no. 10, pp. 457-478, 2019.