摘要
为了解决用单一统计模型进行程序错误定位时存在的泛化能力较弱、错误类型覆盖不广和对测试用例过度依赖等问题,提出了组合统计模型的方法。该方法根据各个弱计算模型的特点,对它们进行了加权组合。组合统计模型具有更好的错误查找性能和抗噪能力,能够有效定位程序中不同类型的错误代码。通过组合模型和单一模型的对比实验结果表明,组合模型有效地提升了系统的分析能力和泛化性能。
To deal with the problems such as weak generalization, poor coverage on bug patterns and strong dependence on test suites that happened in applying unique statistical model in bug localization, a method of composed statistical model is proposed. Composed statistical model assembles several weak models into one system by weights according to their characteristics. This model, which has stronger capability on fault localization and noise tolerance, can find the bugs in various patterns effectively. The improvement on analysis performance and generalization is verified by comparative experiments between composed model and unique model.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第19期4218-4220,4231,共4页
Computer Engineering and Design
基金
国家自然科学基金项目(60673120)
关键词
错误定位
统计分析
组合统计模型
运行时状态
程序测试
fault localization
statistical analysis
composed statistical model
runtime states
program test
