A LEARNING METHODOLOGY FOR A CLASSIFYING MODEL FOR ONTOLOGY ALIGNMENT

Abstract

Ontology alignment is a common and successful way to reduce the semantic heterogeneity among ontologies, relying on the application of similarity functions to decide whether a pair of entities from two input ontologies corresponds to each other. There are several similarity functions proposed in the literature capturing distinct and complementary perspectives, but the challenge is on how to combine their use. This paper presents a methodology to automatically learn a classifier that combines distinct string-based similarity functions for the ontology alignment task, through machine learning. The proposed approach was evaluated experimentally on sixteen scenarios defined on top of the Ontology Alignment Evaluation Initiative (OAEI).Ontology alignment is a common and successful way to reduce the semantic heterogeneity among ontologies, relying on the application of similarity functions to decide whether a pair of entities from two input ontologies corresponds to each other. There are several similarity functions proposed in the literature capturing distinct and complementary perspectives, but the challenge is on how to combine their use. This paper presents a methodology to automatically learn a classifier that combines distinct string-based similarity functions for the ontology alignment task, through machine learning. The proposed approach was evaluated experimentally on sixteen scenarios defined on top of the Ontology Alignment Evaluation Initiative (OAEI).