摘要：AbstractKnowledge graph based explainable recommendation system is a kind of personalized recommendation which uses side information to solve the reason why recommending an item. Previous study has not fully explored the connection between users and items in knowledge graph, especially the problem of overall semantic representation, and can not capture the high level semantic representation of the path, it is difficult for existing path-based methods to clarify the overall semantics of paths. Especially when the path contains similar entities but different relationship. In this paper, we propose a model named Meta-Path-based Explainable Recommendation System (MPERS) to represent the paths in the knowledge graph through the semantic information of entities and relationships, and distinguish the different contributions that different paths make to conduct users’ preference. The experimental study demonstrates the superiority of our method compared with the state-of-the-art ones
