A three learning states Bayesian knowledge tracing model.

This paper proposes a Bayesian knowledge tracing model with three learning states by extending the original two learning states. We divide a learning process into three sections by using an evaluation function for three-way decisions. Advantages of such a trisection over traditional bisection are demonstrated by comparative experiments. We develop a three learning states model based on the trisection of the learning process. We apply the model to a series of comparative experiments with the original model. Qualitative and quantitative analyses of the experimental results indicate the superior performance of the proposed model over the original model in terms of prediction accuracies and related statistical measures.