Automatic and structure-preserved ontology mapping based on exponential random graph model

摘要

Ontology has been widely used as the context representation in ubiquitous environment or smart spaces. However, different ontology representations are adopted in different spaces which exhibit great variation both in the vocabulary and level of detail. In this paper, we propose an automatic and structure preserved ontology mapping method based on exponential random graph model, termed ERGMap. Various representations of the sports ontology are adopted to evaluate the mapping accuracy of ERGMap. Our simulation results show that ERGMap achieves more than 86% of the optimal accuracy when two representations to be mapped are highly related and more than 76% of optimal accuracy when the representations are loosely related. To our best knowledge, ERGMap is the first method proposed, which performs full automatic ontology mapping process and generates a structure-preserved ontology as its output.