Ontology generation through the fusion of partial reuse and relation extraction

Abstract

Ontology generation-a process to automatically create ontologies from existing knowledge sources-has become a key issue with the emergence of the semantic web. Though many researchers are trying to automate this process by exploiting machine learning and data mining techniques, the results remain under exploration. At the same time, when more and more ontologies are available online, it is important to reuse existing ontologies to a certain extent. In this paper, we present a semi-automatic ontology generation system (OntoGenerator) by partially reusing existing ontologies via a modularization technique and a ranking strategy. In order to enrich the semantics of the generated ontology, we integrate natural language-based, non-taxonomic relation extraction into the system. OntoGenerator is aimed at supporting ontology reuse in semantic indexing. Another objective is to evaluate the maturity of the semantic web by applying its technologies in ontology generation. Copyright © 2008, Association for the Advancement of Artificial Intelligence.