The Semantic Gap

Abstract

The automated processing of multimedia resources is challenging, mainly due to the huge gap between what computers can interpret and what humans understand, known as the Semantic Gap. Automatically extractable low-level features, such as dominant color or color distribution, are suitable for a limited range of practical applications only, and are not connected directly to sophisticated human-interpretable, high-level descriptors, which can be added manually. Several attempts have been made in the last decade to bridge the Semantic Gap by mapping semistructured controlled vocabularies to RDFS and OWL. However, these mappings inherited issues from the original XML and XSD vocabularies, some of which can be addressed by combining upper and domain ontologies, rule formalisms, and information fusion.