Multilingual ontology alignment based on visual representations of ontology concepts

Abstract

Image search represents one of the most frequent user actions on the Internet. Existing image search engines do not understand the images they return, nor do they support multilingualism. These issues can be addressed with the introduction of a semantic layer. The semantics is encoded in ontologies, which contain structured information about a domain of application. In order to provide semantic interoperability between (multilingual) ontologies, it is necessary to obtain semantic correspondences - ontology alignments. Various strategies have been proposed for multilingual ontology alignment. In this concept paper, the idea of alignment discovery based on semantic similarity of visual representations of ontology concept is explored. Copyright © IARIA, 2012.