Adaptive Learning in Quran Memorization: A Bayesian Network-Based Approach

Memorizing the Qur'an is a complex learning process that requires consistency, motivation, and effective repetition strategies. Differences in memory capacity and verse difficulty require an adaptive and personalized learning approach. Traditional, uniform methods are not fully able to adapt to the individual needs of students. This study proposes the application of Bayesian Network (BN) as a probabilistic model to support adaptive learning systems in the context of memorizing the Qur'an. The BN model maps the relationship between variables such as Retention Level, Review Frequency, Difficulty Level, Fluency Score, and Error Probability to represent uncertainty and make inferences about the students' memorization conditions. Based on the inference results, the system provides adaptive recommendations in the form of a muroja'ah schedule and the addition of new memorization according to the abilities of each student. Experimental results show that the BN model can increase memorization retention by up to 19.4% and reduce error probability by 37.9% compared to traditional methods. Furthermore, the inference accuracy reached 86% with an F1-score of 0.83, indicating the model's reliability in predicting memorization conditions. BN-based adaptive systems have proven effective in personalizing learning paths, increasing the efficiency of muroja'ah, and optimizing the memorization reinforcement process. These findings suggest that the integration of artificial intelligence in Islamic education can strengthen data-driven learning approaches, as well as open new directions for the development of adaptive, scalable, and sustainable Qur'an learning systems.