Uplifting Relational Data to Knowledge Graphs using RDF*

Abstract

The mapping of relational data into Knowledge Graphs (KGs) is a prospective method to create large-scale KGs. Due to the mapping methods' effectiveness and practicality to uplift relational data into KGs, various mapping approaches such as R2RML have been proposed. The conventional mapping methods are based on the RDF data model. However, RDF reveals some drawbacks in representing complex knowledge structures. A simple syntactic extension of RDF called  RDF* has been proposed to resolve difficult problems in RDF. This paper presents a novel mapping method to transform relational data into RDF* constructs. The proposed method focuses on the decomposition of relational data into RDF* constructs instead of the conventional mapping approaches to configure RDF structures according to the dependency relations of relational data. This paper analyzes the functional properties of RDF* constructs and demonstrates the decomposition of relational data into RDF* constructs with typical examples. In addition, a mapping schema diagram (MSD) and a mapping description language are described in this study. This paper proposes the construction of KGs based on RDF* and creates a new research area concerning mapping relational data into RDF*.