Title: Decision trees in automatic ontology matching

Authors: Siham Amrouch; Sihem Mostefai; Muhammad Fahad

Addresses: Computer Science Department, Faculty of Engineering Sciences, Annaba University, 23000 Annaba, Algeria; Computer Science Department, Faculty of Science and Technology, Souk Ahras University, 41000 Souk Ahras, Algeria ' MISC Laboratory, Université Constantine 2 - Abdelhamid Mehri, Nouvelle ville Ali Mendjeli, BP: 67A, Constantine, Algeria ' Decision & Information Sciences for Production Systems (DISP), CERRAL Center, University of Lyon 2, Bron 69676, Lyon, France

Abstract: The semantic web progress gave rise to a growing number of ontologies in various fields. In order to allow knowledge reuse, ontology matching is an interesting option. In this paper, we propose an ontology matching system that performs class, property and instance matching. This latter is usually achieved by means of Natural Language Processing (NLP) techniques, which are context dependent. To avoid the limits of NLP, we use a decision tree-based instance matching scheme. Decision tree is one of the most widely used learning algorithms for inductive inference and classification. Our system works on OWL-DL ontologies, that is ontologies expressed in the Description Logic version of the Ontology Web Language. This ensures the maximum of expressiveness, completeness and decidability. Our approach is tested with the benchmark and conference tracks of the OAEI'2015 campaign. It shows very promising results since it outperforms other matching systems in most of the test cases.

Keywords: ontology matching; instance matching; OWL-DL; linguistic similarity; wordNet; decision trees; Weka; semantic web; knowledge reuse.

DOI: 10.1504/IJMSO.2016.081585

International Journal of Metadata, Semantics and Ontologies, 2016 Vol.11 No.3, pp.180 - 190

Received: 05 Apr 2016
Accepted: 13 Oct 2016

Published online: 16 Jan 2017 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article