Empirical Evaluation of Spectrum-Based Fault Localization (SBFL) Technique in Software Fault Localization
Nadhratunnaim Nasarudin, Richard F. Paige, Habiel Zakariah, Shuzlina Abdul-Rahman · International Journal of Research and Innovation in Social Science · 2024
Debugging is a key activity in the software development process. It has been used extensively by developers to attempt to localize faults, while enhancing the quality and performance of software. There has been a significant amount of study in developing and enhancing fault localization techniques, which are used in assisting developers to locate faults within a body of code. An experiment has been carried out to evaluate the accuracy and execution time of the Spectrum-Based Fault Localization (SBFL) technique. SBFL is generally argued to be the most effective (accurate, fastest); amongst the family of SBFL techniques, where Ochiai, one of SBFL formulas to calculate suspiciousness, has been shown to provide the best performance against all metrics. This paper presents an empirical evaluation of SBFL techniques in the context of software fault localization. Using the Defects4J dataset with 395 faults and bug reports, the study investigates the accuracy and performance of this technique, to provide insights into their effectiveness in real-world debugging scenarios. The result from the experiment shows that the SBFL accuracy for the Top 1 is 47.6%, the Top 5 is 59.56% and the Top 10 is 64.56% with 1.74s Runtime per bug(s), where time spent on each bug in localizing fault.