Coverage-Aware Contribution-based Test Case Reduction for Mutation-based Fault Localization

MBFL (Mutation-Based Fault Localization) is a powerful approach for software debugging but suffers from high costs due to exhaustive mutant executions. While existing strategies such as Contribution-Based Test Case Reduction (CBTCR) enhance efficiency through selective prioritization of high-contribution test cases, prior research neglects structural similarities between passed and failed test cases. This paper proposes CACBTCR, a Coverage-Aware Contribution-Based strategy for MBFL Test Case Reduction. CACBTCR calculates the structural similarity of passed tests and failed tests using coverage similarity and then integrates coverage similarity with contribution values to prioritize passed tests that are structurally closer to failed ones. The experiment evaluated 383 real faults from the Defects4J benchmark, and CACBTCR outperforms other reduction strategies of MBFL (IETCR, FTMES, and CBTCR) at the metric of Top-N and MAP. Furthermore, CACBTCR achieves 85.43% execution cost reduction while maintaining fault localization effectiveness comparable to original MBFL techniques.