Explainable Educational Recommender Systems: A Causal Graph Approach for Predicting and Enhancing Student Performance
T V Sathyanarayana Research Scholar Shri Jagdishprasad Jhabarmal Tibrewala University tv_sathya@yahoo.co.uk
Dr. ARCHANA TUKARAM BHISE Research Guide Shri Jagdishprasad Jhabarmal Tibrewala University archanab34@rediffmail.com
Abstract :
This paper proposes an explainable educational recommender system that leverages causal graphs to model the complex relationships between learning activities, student behaviors, and performance outcomes. Unlike correlation-based approaches, our framework incorporates causal inference to identify actionable learning pathways, ensuring that recommendations are not only predictive but also pedagogically meaningful. By integrating explainability mechanisms, the system provides both students and educators with transparent justifications for each recommendation, highlighting the causal factors influencing predicted performance. Experiments conducted on real-world educational datasets demonstrate that the proposed approach improves both prediction accuracy and interpretability, while offering actionable insights for personalized learning interventions. This work contributes to bridging the gap between predictive accuracy and explainability in educational recommender systems, ultimately fostering more trustworthy and effective technology-enhanced learning environments.
Keywords: Explainable AI, Educational Recommender Systems, Causal Graphs, Student Performance Prediction, Personalized Learning, Interpretability, Causal Inference
