Interpretable MOOC Course Recommendations Based on Reinforcement Knowledge Graph
Yifan Zhang, Yiwen Zhang, Xiao‐Lan Cao, Yuchengqing Yu · 2024
With the large amount of educational information on online education platforms causing information overload, it is particularly important to select personalized content suitable for users from a large number of courses. This paper proposes a reinforcement learning method using MOOC knowledge graph as the environment to generate explainable recommended course paths, recommend courses to users and give clear explanations, so that students can understand the reasons and logic of the recommendations. The MOOC knowledge graph covers a variety of complex relationships and learners' historical records, while reinforcement learning simulates the interaction between users and courses to make recommendation decisions. Experiments were conducted on the MOOC data set, and the experimental results proved that the method combining knowledge graph and reinforcement learning proposed in this article improved the NDCG and HR indicators.