HMBFL: Higher-Order Mutation-Based Fault Localization
Zheng Li, Butian Shi, Haifeng Wang, Yong Liu, Xiang Chen · 2021 8th International Conference on Dependable Systems and Their Applications (DSA) · 2021
Fault localization is one of the most important processes in debugging. Among various automated fault localization techniques, Mutation-Based Fault Localization (MBFL) is one of the commonly studied that could achieve better performance in single-fault localization scenarios. However, the MBFL technique adopted First-Order-Mutants (FOMs) does not perform well in multiple-fault localization. In this paper, we propose an approach HMBFL (Higher-Order Mutation-Based Fault Localization) that first applies Higher-Order-Mutants (HOMs) in fault localization. To utilize HOMs on fault localization, we present three methods for calculating statements' suspiciousness (i.e., Averaging, Maxing, and Frequency). Furthermore, to generate more effective HOMs for fault localization, we propose two strategies that based on the traditional techniques (i.e., SBFL-guided and MBFL-guided). Our empirical results on 112 real-world multiple-fault programs from Codeflaws show that HMBFL outperforms SBFL techniques and traditional MBFL techniques at the metric of EXAM and HMBFL place more faults at the top 1, 3, 5 ranks.