Recommendation Algorithms Based on Knowledge Graphs and Graph Attention Networks

Recommender systems are essential for delivering tailored content and services to users, as a key means to solve the information overload problem. However, it also encounters challenges like data sparsity and the cold start problem. In order to cope with these challenges, the knowledge graph is used as auxiliary information with graph neural network-based recommendation algorithm, an innovative recommendation algorithm, KGAE (knowledge graph attention recommendation of transe). First, a collaborative knowledge graph (CKG) is constructed by leveraging both the user-item interaction graph and the item knowledge graph, then the translation distance model TransE is embedded into the graph attention network, The GraphSAGE technique is employed to refine target node representations, utilizing an attention mechanism to discern neighbor importance and thereby enhance recommendation accuracy. Comparison experiments are conducted with five mainstream recommendation algorithms on two public datasets. Comparison experiments are conducted with five mainstream recommendation algorithms on two public datasets, and the results show that the proposed method can significantly improve the recommendation effect.