Temporal Bayesian Knowledge Tracing: Integrating Time-Based Features for Quran Memorization Pathways

Temporal Bayesian Knowledge Tracing (TBKT) is a knowledge prediction model designed to address the challenges of personalizing Quran memorization pathways. This model extends the Bayesian Knowledge Tracing (BKT) approach by integrating time-based features, such as time gaps and repetition frequency, to more accurately model the dynamics of learning and forgetting. By considering temporal factors, TBKT is able to adjust the transition probabilities between knowledge states (learning and forgetting) based on the time pattern and individual learning behavior. This study evaluates TBKT using a Quran memorization log dataset that includes learning sessions, time between sessions, repetition frequency, and recall performance. The evaluation results show that TBKT has an accuracy of 0.85, a precision of 0.833, and an F1-score of 0.909, outperforming the traditional BKT model. Further analysis reveals that repetition frequency has a positive effect on increasing the probability of learning, while long time between sessions tends to increase the probability of forgetting. TBKT provides a more adaptive and effective approach in predicting retention and personalizing Quran memorization strategies. This model can be used by educators to optimize memorization repetition schedules, reduce the effects of forgetting, and increase learners long-term retention