Debugging Effectiveness of LBT: An Empirical Study

Abstract

Debugging is an essential step in the software development life-cycle. Automatic test generation techniques and tools have significantly improved software development. Test generation tools are evaluated based on their ability to reveal faults or coverage capability. The Learning-based Testing (LBT) is a black-box testing tool that uses a learning algorithm to generate tests. This tool has not been explored concerning debugging effectiveness. We conducted a study involving 20 human subjects to evaluate LBT and its effectiveness in debugging. We used two case studies of reactive systems and two open-source model-based testing tools to evaluate the effectiveness of LBT over other tools. We have discovered that LBT is an effective tool for automatic test case generation and show how LBT may improve the debugging process. The study found that LBT can aid less experienced and experienced developers with debugging. The effectiveness of LBT is 95%, and the efficiency is higher than other tools. Besides, we discover that skilled developers have performed satisfactorily on other tools, yet there is a noticeable difference in time.