一种基于模型结合的错误定位方法.docVIP

 n 一种基于模型结合的错误定位方法 唐启锋 1,2 ，许蕾 1,2，钱巨 3，陈林 1,2，张震宇 4 (1. 南京大学计算机软件新技术国家重点实验室，南京 210093； 2. 南京大学计算机科学与技术系，南京 210093； 3. 南京航空航天大学计算机科学与技术学院，南京 210016； 4. 中国科学院软件研究所，计算机科学国家重点实验室，北京 100080) 摘要：目前大多数错误定位技术的研究均基于单一类型的程序节点（如语句、谓词等），其 效果往往只在定位相应类型的错误时表现较好，而定位其他类型的程序错误时则表现不佳。 为此，借鉴机器学习领域中集成学习的思想，建立多错误定位方法相结合的错误定位模型， 并综合了基于语句覆盖信息和程序谓词信息这 2 种错误定位方法，提出了 3 种新的错误定位 方法。实验结果表明，相对于此前单一的方法，所提出的 2 种方法具有更高的错误定位效率 和更强的适应性。 关键词：错误定位；集成学习；程序分析；软件调试 中图分类号：TP311.5 文献标志码：A A fault localization method based on model combination Tang Qifeng1,2, Xu Lei1,2, Qian Ju3, Chen Lin1,2, Zhang Zhenyu4 (1.State Key Laboratory of Novel Software Technology, Nanjing University, Nanjing 210093, China；2.Department of Computer Science and Technology, Nanjing University, Nanjing 210093, China；3.College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China；4.State Key Laboratory of Computer Science, Institute of Software, Chinese Academy of Sciences, Beijing 100080, China) Abstract: Automated fault-localization technique has become a hot topic in research fields of software engineering. However, most of techniques are based on single type program node(e.g. statement, predicate, etc.) until now, when these techniques utilized to locate the corresponding types of errors, it has good performance in generally. But in other situation, the opposite is true. Therefore, we draws on the method of Integrated Learning in machine learning to contribute a fault-localization model which combines the information of statement-based coverage and predicate-based coverage method. In addition, three new fault-localization methods are proposed. The experimental results show that our methods have better effectiveness and adaptability for fault-localization than previous methods. Key words: fault localization; ensemble learning; program analysis; debugging 1 引言 提高软件质量是软件工程亟需解决的重大问题之一。软件调试作为一种提高软件质量的 重要手段而备受重视。传统的调试方法主要包括设置断点，重复运行程序和检查相应的程序 状态点。但是，这些调试方法通常会耗费大量的时间和金钱，并且已经