A New Similarity Measure for Instance Data Matching
Kamran Zamanifar, Farinaz Alamiyan, Islamic Azad, University-Najaf Abad, Najaf Abad · 2011
Abstract. Instance based ontology matching is a new approach which uses the extensions of concepts to solve matching problems. Instance matching is the task of recognizing similarity between instances with different ontology descriptions for the same real world objects. It is crucial in this ontology matching area like other data integration issues, and the quality of its algorithm can extremely affect the performance of this way of matching because it can change the overlap of extensional information; that is, common instances of concepts among different ontologies. In this paper, we propose a distributional instance similarity measure and study how the choice of this measure improves the quality of instance mapping in ontology matching scenario.