Comprehensive approach to semantic similarity for rapid data integration

Abstract

It is greatly significant to combine data with diverse structures and supply a unified view of these data for the user, especially for sharing data residing in heterogeneous data sources via the internet. This paper introduces a fast and novel method to data integration among different systems, which is based on ontology similarity in a language agnostic way. The fundamental ontological entities are extracted from multiple data sources according to the same mapping rules. By means of the improved edit distance algorithm, similarity measurement consists of determining the levels of similarity among the ontological entities for aiding in the construction of data integration platform. A web-service based architecture is presented along with the set of layers designed to achieve rapid data integration from different aspects such as data center, ontology extraction and similarity measurement, which aims to make this architecture more flexible. The prototype implemented by the proposed approach shows satisfying results against other techniques.