A Recommendation Method for Electronic Components Based on Knowledge Graph

Enterprises are facing significant challenges due to electronic component shortages and the need for domestic alternatives. In this paper, we propose the integration of knowledge graphs and recommender system technologies to address these issues. First, the recommended model in this paper adopts the idea of alternating learning to treat recommendation task and knowledge graph embedding task as two relatively independent modules, and designs an information fusion unit to fuse user behavior information in recommendation task with entity structure information in knowledge graph embedding task. Second, the Word2Vec word vector model was used to extract electronic component characteristic word vectors as part of the initialization of electronic component entity nodes in the knowledge graph embedding task, introducing semantic features to the model. Finally, the proposed method achieves an AUC of 91.5% and an ACC of 85.8% in electronic component recommendation. The experimental results indicate that the method is feasible.