Document Type : Other
Author
Department of Computer Engineering, Ferdowsi University of Mashhad, Bahonar Street, Mashhad, Khorasan Razavi ,Iran
Abstract
Spectrum-based fault localization (SBFL) is a widely used technique that utilizes coverage data and test outcomes to calculate a suspiciousness score for each program statement. The fundamental hypothesis of SBFL is that a statement covered by more failed test cases and fewer passed test cases is more likely to be faulty. However, the effectiveness of SBFL is hindered by coincidental correctness, which occurs when a fault is executed but no failure is detected. Additionally, traditional SBFL methods assign equal weight to all failed tests, despite some failed tests containing more valuable information. This study aims to enhance SBFL performance by employing a fuzzy expert system to address these challenges. Thirteen open-source subject programs were used to evaluate the efficiency of the proposed FSBFL method. Experimental results, assessed using four key metrics, demonstrate that FSBFL outperforms popular spectrum-based fault localization techniques.
Keywords
Main Subjects
)