Enriching Ontologies with Multilingual Information

Abstract

Organizations working in a multilingual environment demand multilingual ontologies. To solve this problem we propose LabelTranslator, a system that automatically localizes ontologies. Ontology localization consists of adapting an ontology to a concrete language and cultural community.
LabelTranslator takes as input an ontology whose labels are described in a source natural language and obtains the most probable translation into a target natural language of each ontology label. Our main contribution is the automatization of this process which reduces human efforts to localize an ontology manually. First, our system uses a translation service which obtains automatic translations of each ontology label (name of an ontology term) from/into English, German, or Spanish by consulting different linguistic resources such as lexical databases, bilingual dictionaries, and terminologies. Second, a ranking method is used to sort each ontology label according to similarity with its lexical and semantic context.
The experiments performed in order to evaluate the quality of translation show that our approach is a good approximation to automatically enrich an ontology with multilingual information.