Relation Modeling on Knowledge Graph for Interoperability in Recommender Systems

Seungjoo Lee, Seokho Ahn, Young-Duk Seo

Abstract: Network technology has changed how people consume content through various channels, which constantly generate a large amount of data. To effectively utilize this data, many researchers have focused on interoperability, or the ability to use information from multiple systems together. However, in the field of recommender systems, few studies have considered interoperability. Existing methods for guaranteeing interoperability in recommender systems have limitations in their ability to model low-order relationships for data integration. In particular, there has been no study that ensures interoperability for knowledge graph-based recommender systems, which are suitable structures for integrating heterogeneous data. Therefore, we propose an integration method for multiple systems optimized for knowledge graph-based learning. This method can extend the knowledge graph through deep learning-based relation modeling of entities and ensure interoperability for the recommendation system. Our experimental results confirm that this method improves the performance of existing recommendation algorithms.