OtO matching system: A multi-strategy approach to instance matching

Abstract

In this paper we describe an ontology to ontology (OtO) matching system which implements a novel instance matching algorithm. The proposed multi-strategy matching system is domain independent and fully customizable at any level. It optimizes the instance matching process by leveraging (a) the rich semantic knowledge from the schema matching results, (b) the implicit knowledge of the domain expert by capturing the identification power of the properties and (c) the probability estimation of the result's validity, in order to accurately detect the ontology instances that represent the same real-world entity. Furthermore we evaluate the system with the ISLab Instance Matching Benchmark in the Ontology Alignment Evaluation Initiative 2009 campaign and report the results. © 2012 Springer-Verlag Berlin Heidelberg.