A Maximum Semantic Reservation Mapping Method Based on Ontology-to-graph Database

Abstract

Ontology is a core concept model in a knowledge graph which describes knowledge in the form of a graph. With the increase in knowledge graphs, the semantic relationships between concepts become more and more complex, which increases the difficulty of reserving its semantic integrity when storing it in a database. In this paper, we propose an ontology-to-graph database mapping method, which can reserve maximum semantic integrity and reduce redundant information simultaneously with high storage efficiency and query efficiency. In detail, the mapping method uses an anonymous class storage strategy to handle indefinite long nested structures, a multivariate functional relation storage strategy for multivariate semantic analysis, and an SWRL (Semantic Web Rule Language) storage strategy for disassembling inference structures. We develop an ontology-to-graph database prototype Neo4J4Onto to implement the mapping method. Experimental results show that our method achieves the maximum semantic integrity with the lowest complexity compared to the 6 baseline methods. Besides, compared to graphDB, Neo4J4Onto has better storage and query efficiency, and the concept models retrieved by Neo4J4Onto are more complete.