Optimizing learning paths to improve learning outcomes and learner engagement has always been a challenge in the field of personalized learning and online education. Traditional recommendation systems often suffer from limitations such as data sparsity and poor interpretability, which restrict the effectiveness of personalized recommendations. To address these issues, this paper proposes a novel course recommendation model—Reinforced Heterogeneous Knowledge Graph Reasoning for Course Recommendation (RHCR). Specifically, RHCR introduces a heterogeneous course knowledge graph to mitigate issues like sparse data and weak interactions, and formulates course path reasoning as a Markov Decision Process (MDP). By utilizing the Asynchronous Advantage Actor-Critic (A3C) algorithm enhanced with Multi-Head Attention and Bidirectional Long Short-Term Memory (MHA-BiLSTM), the model optimizes recommendation paths based on learners’ profiles and historical course data. Experimental results show that RHCR increases Normalized Discounted Cumulative Gain (NDCG) by 8.16% and the Precision by 16.29% on the same dataset, outperforming traditional neural network–based methods. Moreover, it alleviates data sparsity and improves recommendation interpretability, providing an effective solution for personalized learning path optimization.
Li et al. (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: