Improving the performance and explainability of knowledge tracing via Markov blanket

Bo Jiang, Yuang Wei, Ting Zhang, Wei Zhang

Abstract: Highlights•Considering knowledge tracking from the perspective of causal discovery for the first time.•Using causal relationships can enhance the interpretability and performance of the knowledge tracing model.•The performance of the knowledge tracing model based on Markov blankets on public datasets is better than other models.•Knowledge tracing model based on Markov blankets contributes to the advancement and development of sustainable education.