The recommendation system can help users filter out the items that meet their needs among the massive resources, and theevolving recommendation system provides new?ideas for online education. As an important part of online education, online courseresource recommendation currently has the problems of overload of course resources and?
lack of interpretability of course recommendationresults. In this regard,we propose an online course recommendation model based on knowledge subgraph and attention mechanism to useknowledge subgraph for recommendation. Different from the model that directly uses the knowledge graph for recommendation andignores the problem of inaccurate knowledge representation,the proposed model first uses the Node2vec random walk method to extractthe connected subgraph connecting user-course pairs from?
the knowledge graph,and then encodes the subgraph through the hierarchicalattention network to generate a subgraph embedding for the prediction of the courses required?
by the user. Finally,a list of Top-N recommended courses is generated,and an interpretability description of the proposed model is given. In order to verify the effectiveness of theproposed model,the data set on the " MOOC ( MOOC) of Chinese universities" was used as the sample,and the experimental resultsshow that compared with the KGCN-PN,GAT,KGAT and POCR,the proposed model improves the NDCG,HR and MRR evaluationindexes by 10. 6% ,9. 41% and 13. 7% ,respectively.