Personalized Learning Path Recommendation for E-Learning Based on Knowledge Graph and Graph Convolutional Network
Xiaoming Zhang, Shan Liu, Huiyong Wang
Abstract
In e-learning, the increasing number of learning resources makes it difficult for learners to find suitable learning resources. In addition, learners may have different preferences and cognitive abilities for learning resources, where differences in learners’ cognitive abilities will lead to different importance of learning resources. Therefore, recommending personalized learning paths for learners has become a research hotspot. Considering learners’ preferences and the importance of learning resources, this paper proposes a learning path recommendation algorithm based on knowledge graph. We construct a multi-dimensional courses knowledge graph in computer field (MCCKG), and then propose a method based on graph convolutional network for modeling high-order correlations on the knowledge graph to more accurately capture learners’ preferences. Furthermore, the importance of learning resources is calculated by using the characteristics of learning resources in the MCCKG and learners’ characteristics. Finally, by weighting the two factors of learners’ preferences and the importance of learning resources, we recommend the optimal learning path for learners. Our method is evaluated from the aspects of learner’s satisfaction, algorithm effectiveness, etc. The experimental results show that the method proposed in this paper can recommend a personalized learning path to satisfy the needs of learners, thus reducing the workload of manually planning learning paths.