Layered Mixing Miner: A Process Mining Algorithm for Complex Programming Debugging Data

In the context of increasingly complex application scenarios and unstructured debugging data in software engineering education, analyzing students' programming and debugging processes presents significant challenges. This study introduces the Layered Mixed Miner (LMM), a novel process mining algorithm designed to address these challenges by generating inter-pretable models from complex and unstructured student programming and debugging data. The algorithm integrates various process mining techniques by constructing an initial process tree, calculating filtering thresholds, and employing diverse methods to construct process models based on different filtering results. Consistency tests demonstrate that LMM achieves higher accuracy and precision compared to established benchmarks, providing a reliable and insightful representation of debugging behaviors. Moreover, expert feedback further confirms the algorithm's superior interpretability and usability over baseline methods.