Abstract: Personalized learning resource recommendation represents an advanced form of deep integration between technology and education,which has received widespread attention from academia in recent years.Existing reviews have provided valuable discussions on basic theoretical frameworks,recommendation methods,and evaluation metrics.Building upon this foundation,the classification system and algorithmic analysis of personalized learning resource recommendations can be further enriched,especially through a comprehensive review of traditional recommendation methods and research on the application of emerging large language models in personalized learning recommendations.This review constructs a multi-dimensional classification framework,systematically categorizing personalized learning resource recommendations from three dimensions:application scenarios,algorithm categories,and learner characteristic modeling.It comprehensively explains the technical principles and latest developments of traditional recommendation algorithms,knowledge-based algorithms,machine learning algorithms,intelligent optimization algorithms,deep learning algorithms,and reinforcement learning algorithms.The review systematically analyzes the technical challenges facing personalized learning resource recommendations and the dilemmas from the learners’ perspective,and accordingly proposes future research directions such as integrating multiple recommendation methods,introducing educational theory guidance,and improving data quality.This review aims to provide educational technology researchers and practitioners with a systematic theoretical framework and technical roadmap to promote the continuous optimization and innovative development of personalized learning ecosystems.

Key words: Smart education, Personalized learning, Personalized resource recommendation, Recommendation algorithms, Machine learning, Intelligent optimization