Improving Mutation-Based Fault Localization with Plausible-code Generating Mutation Operators

This paper proposes a new mutation operator using neural network to generate plausible code elements to improve performance of mutation-based fault localization on omission faults. Unlike the existing mutation operators, the proposed mutation operator synthesizes new code elements at a given mutation site with a neural language model. We extended MUSE to use the proposed mutation operator, and conducted a case study with 3 omission faults found in JFreeChart of Defects4J. As a result, the accuracy of MUSE with the new mutation operator increased significantly in all three faults.