Knowledge Graph Modeling based on Generalized Property Graphs

Abstract

As a popular emerging technology, knowledge graphs (KGs) have become a platform of knowledge applications and services. Two dominant graph-based knowledge models, RDF and LPG, are widely used to construct large-scale KGs. It is argued that these models have some limitations to cope with complicated knowledge structures. This paper proposes a noble generalized property graph model that can seamlessly realize the various types of knowledge structures with compact expressiveness and robust formalism. The paper also describes several structures of knowledge graphs to demonstrate the expressive capability of the proposed property graph model. Since the generalized graph model is compatible with RDF and LPG, it can be practically applied to the usual KGs with effective performance.