Learning Preference Recommendation with Heterogeneous Graph Neural Networks in MOOC
Shuyan Wang, Yaping Li · 2021 4th International Conference on Artificial Intelligence and Pattern Recognition · 2021
In order to solve the problem of low utilization of MOOC resources and information overload, in this paper we propose a MOOC recommendation model based on heterogeneous graph neural networks combining with attention mechanism. First, we design multiple meta-paths containing semantic relations to guide the propagation of users’ preferences in the graph neural network based on residual connection. And then combine with the attention mechanism to strengthen the factors that effectively affect users’ preferences. Finally, we use regularized matrix decomposition to predict the rating. In this study, we evaluate the proposed model by collecting and organizing the MOOC dataset. Through repeating experiments, we prove the evaluation metrics of HGMCRec is better than other baseline methods. And we also prove the usability of this model on two public datasets.