A multi-stage strategy for ontology mapping resolution

Ontology mapping is a challenge we are facing in information integration, retrieval and sharing. In this paper we present a multi-stage way which considers all the available information to improve the accuracy of ontology mapping. The initial similarities of concepts of two ontologies are obtained at the first stage based on the information of labels, instances and properties. Then former indexed experiences that imbed domain knowledge are used to modify the result. After that we use a preprocessing process to modify the structures of ontologies to reduce the heterogeneity of the structures. Then a graph-based iteration process is carried out, which considers the similarity of structures. Finally the correspondences of properties of the equivalent concepts are identified based on the labels, types, values and axioms, wherein the axioms may be predefined or may be discovered via data mining from instances.