An Artifact-Centric Process Mining Approach for Learning Style Analytics

As an integrated discipline encompassing data mining, machine learning, process modeling and analytics, process mining is increasingly being applied in the field of education and has emerged as a prominent research topic. Traditional business process modeling approaches, which are primarily based on control flow rather than data flow, exhibit a limited capacity to capture a holistic view of critical business data within complex business procedures. This study focuses on the impact of data-driven process modeling techniques on the performance of analytical models and proposes an artifact-centric process mining approach for learning style analysis. Based on the artifact life-cycle model, we extracted sequences of data attribute operations that encapsulate learning style features. The similarity among different data attribute operation sequences was quantified. The proposed method was evaluated using the OULAD, a benchmark dataset in the learning analytics domain. Experimental results demonstrate that the method effectively enhances the performance of learning style prediction models, with SVM and GBoost algorithms outperforming other modeling approaches.