A Personalized Recommender System based-on Knowledge Graph Embeddings

Abstract. Knowledge graphs have proven to be effective for modeling entities and their relationships through the use of ontologies. The recent emergence in interest for using knowledge graphs as a form of information modeling has led to their increased adoption in recommender systems. By incorporating users and items into the knowledge graph, these systems can better capture the implicit connections between them and provide more accurate recommendations. In this paper, we investigate and propose the construction of a personalized recommender system via knowledge graphs embedding applied to the vehicle purchase/sale domain. The results of our experimentation demonstrate the efficacy of the proposed method in providing relevant recommendations that are consistent with individual users.

Introduction. In today’s world, where there is an abundance of information, recommender systems have become increasingly important in curating and presenting relevant content to users. These systems are designed to assist individuals by providing personalized suggestions and recommendations, which makes information discovery and decision making more efficient. The use of knowledge knowledge graphs for recommender system has demonstrated the ability to produce highly accurate recommendations that are also straightforward to interpret and explain [14, 24]. The essential elements of knowledge graphs are to enhance the organization and structure of information which is able to effectively define a measure of relatedness between entities [6]. A knowledge graph by means of ontology creates a structured framework for a set of concepts or terms within a specific domain by arranging them in a hierarchical manner, and by using relation descriptors to model the connections between these concepts or terms. This provides a standardized lexicon for representing entities within that domain [11, 16].

Discussion / Conclusion. In this paper, we investigate the construction of an approach for a personalized recommender system based on building knowledge graph embedding, which allows for the extraction and addition of more information into the learningto-rank process. From our work, we illustrate how we can separate building sub-graphs for each relation type in the knowledge graph and use it to build embeddings. The results obtained from the experiment show that our approach is of interest as it provides better results compared to other approaches on our dataset. In our future work, we intend to research the exploitation of ways to explain the recommendation results to a given user by analyzing features learned from knowledge graphs.