A Multi-Perspective Recommendation Model Based on Knowledge Graph

A knowledge graph is a graph structure that represents and organizes knowledge, where nodes represent entities and edges represent the relationship between entities. The model uses the rich information in the knowledge graph, including entity properties, relationships, and semantic similarity, to improve the recommendation effect. However, the traditional knowledge graph is difficult to mine the user intention, so the recommendation effect is limited. In order to solve this problem, the MKGN model proposed in this paper transmits the collaboration signal and knowledge signal through the collaborative knowledge perception module, and then obtains the user and the project embedding through the knowledge attention mechanism; introduces the user interest into the potential vector of the interest module, builds the user-interest-item triplet, obtains the user and combines the user and the project embedding together for the final recommendation score. The model is compared with the baseline algorithm on two public datasets, and the model has better performance than the baseline algorithm model.