基于数据流分析的警报关联研究
larm Clustering based on Data Flow Analysis
在大型软件工程测试中，静态缺陷检测工具产生的警报数量巨大，开发人员需要耗费大量精力来判断警报实际是缺陷还是误报。本文通过在缺陷检测过程中，发掘警报间依赖关系，对警报进行分类；对于每类警报，通过前向数据流分析和后向数据流分析选取主导警报。如果该类中的主导警报为误报，则该类其它警报也为误报。实验表明，本文的聚类方法可以减少40%-50%的人工审查工作。
Static analysis tools have been successfully adopted in softaware testing; however millions of alarms that are generated by the tools are reviewed manually at a very low speed. To help with this review process, we first clustered alarms by discovering dependencies among them such that if the leader alarm of a cluster is a false positive then it is assured that all others in the same cluster are also false. Empirical results show that our clustering methods could reduce 40-50% manual review efforts.
金大海、薛伟、李蕊彤
计算技术、计算机技术
软件工程静态缺陷检测数据流分析缺陷关联?????
Software EngineeringStatic AnalysisData Flow AnalysisAlarm clustering
金大海,薛伟,李蕊彤.基于数据流分析的警报关联研究[EB/OL].(2014-01-07)[2025-08-11].http://www.paper.edu.cn/releasepaper/content/201401-321.点此复制
评论