关键词: 动态切片, 程序调试, 概率推演, 自动化程序分析, 错误定位

Abstract: In order to increase the productivity of the debugging process, this paper proposes a novel strategy for prioritizing dynamic program slices by automatically calculating the probability of correctness of each statement based on the dynamic slices. First, a runtime dependence graph of the observed program outcome is extracted. The next step is the transformation from the dependence network to a Bayesian network. Finally, run a probabilistic inference for the likelihood of correctness of the execution instances and estimates the corresponding correctness of the static statements. Programmers can administer a guided bug locating process using this ranking of correctness belief. The results shows that on average our tool rank the faults to 20.2% of the dynamic slices.

Key words: dynamic slicing, debugging, probabilistic inference, automated program analysis, fault localization