Cluster optimization-based collaborative filtering recommendation algorithm for educational resources under AHP-entropy weight coupling

This study proposes a collaborative filtering recommendation model that integrates hybrid empowerment and clustering optimization to address the challenges of cold start and data sparsity in personalized recommendation on online education platforms. Based on the four factors of task completion rate, learning time, resource preference and test score, we combine the Analytic Hierarchy Process (AHP) and entropy weighting method to construct a subjective-objective balanced weighting system, and design a two-phase clustering optimization strategy: first, we use K-means to classify four types of learning communities to improve data density, and then we use the Item-CF algorithm to achieve efficient recommendation. Experiments show that the model optimizes the Normalized Depreciation Cumulative Gain (NDCG) metrics, and the average recommendation accuracy reaches 0.982, which is both accurate and scalable. By balancing interpretability and efficiency through hybrid assignment, clustering-driven similarity modeling breaks through the cold- start limitation and provides a new solution for personalized recommendation on online education platforms. This study integrates multi-criteria decision-making and computational optimization techniques to promote personalized learning in smart education, improve the adaptation of educational resources and learning efficiency, provide theoretical and practical support for the intelligent transformation of educational services, and help improve educational equity and quality.