StasogFL: A Fault Localization Method Based on Statement Association Graph
Yingxian Guo, Shihai Wang, Dong Un An, Yu Luo · 2024
In the realm of software development, Software Fault Localization (SFL) is critical for enhancing software quality and maintenance. Traditional methods like Spectrum-Based Fault Localization (SBFL) have been foundational but often fall short in handling the complexity and intricacies of modern software systems. This paper introduces a novel graph-based fault localization method, StasogFL, which leverages a statement association graph to capture the intricate relationships within software structures. By integrating control flows, data flows, and function calls into a comprehensive graph model, our approach significantly improves fault localization's precision and efficiency. The empirical results from applying StasogFL on large Java projects demonstrate substantial improvements over traditional SBFL methods and other graph-based approaches, particularly in terms of fault prediction accuracy and localization efficiency. This paper not only showcases the effectiveness of StasogFL but also paves the way for future research to further refine and expand its application across more varied software maintenance scenarios.