An Empirical Study on Spectrum-Based Fault Localization for Student Programs
Yuxing Liu, Zhanwen Zhang, Xuchuan Zhou, Wentao Liu · 2023
Programming competitions have become increasingly prominent in universities, fostering innovation and skill development in computer technology. However, programming faults, an inherent aspect of the development process, can hinder participants' performance and affect the functionality of their solutions. Consequently, it is crucial to establish efficient fault localization techniques to facilitate improved algorithm learning and debugging practices. We investigate the applicability of Spectrum-based Fault Localization (SBFL) in student courses through an empirical study of 122 real student programs gathered from a Chinese university's online testing platform. Additionally, we designed a user study for subjective evaluation. The experimental findings indicate that SBFL successfully localizes faults in student assignments, yielding maximum TOP-1, TOP-3, and TOP-5 of 22, 36, and 61, respectively. Furthermore, the user study reveals that the method is user-friendly, considerably reducing debugging time by over 50% and enhancing debugging efficiency.