【摘要】 大型软件的调试往往需要花费大量人力和时间,因此自动化的软件故障定位方法成为更好的选择.现有故障定位方法大都基于用例的覆盖信息进行分析,没有完全考虑测试用例执行时的顺序信息.针对上述问题,提出测试用例执行轨迹分析的故障定位方法,该方法首先通过控制流分析获得测试用例的执行轨迹,然后将执行轨迹编码成序列信息用于训练循环神经网络,最后将一组虚拟测试用例输入网络中预测代码块的可疑度.通过在Siemens套件上的实验和分析,验证了所提方法的有效性.更多还原

【Abstract】 Debugging large software always takes a lot of manpower and time,so automated software fault localization method becomes a better choice. Most existing fault localization methods are based on the coverage information,yet the sequence information does not be fully considered during testing. To solve the above problem,a fault localization method based on analyzing the execution trajectories of test cases is proposed in this paper. Firstly,the execution trajectory of each test case is obtained through control flowanalysis,and then the execution trajectories are encoded into sequence information to train the recurrent neural networks. Finally,a set of virtual test cases are input into the trained network to compute the suspicious of each basic block. Through the experiment on the Siemens suite,the proposed method proved to be effective.更多还原