A Chinese Movie Recommendation Algorithm Incorporating Knowledge Graph and GNN

To address the issues of poor recommendation effectiveness, data sparsity, and cold start problems in Chinese movie recommendation systems, a Chinese movie recommendation algorithm called TransH-SAGE algorithm has been proposed. This algorithm combines knowledge graph with graph neural networks to embed movie entities, integrating semantic similarity and rating similarity, sorting the similarities, and recommending highly similar movies. Experimental results show that the recall and precision of the recommendations reach 87.22% and 87.14% respectively.