One of the challenging issues in the field of multimedia analysis is to extract high-level semantics from a document. As opposed to the domain of analysis and indexing of textual documents, the visual domain has to face the important challenge of matching human interpretations of image information with the numerical image signature derivable by a computer, defined as the semantic gap problem. Bridging the semantic gap has spurred continued interest recently, in particular concerning the explicit representations of a priori knowledge. In many domains, ontologies are gradually being accepted as the key technology to describe the semantics of information. In the multimedia domain, they have shown to be a promising solution for making semantics explicit.
As a consequence, many multimedia ontologies have been built either to assist multimedia search and retrieval or to navigate in large multimedia collections. These ontologies are usually only hierarchically structured multimedia concept lexicons (or thesauri), rarely used as formal models that support different kinds of reasoning (through description logics) or engineering operations (ontology building, ontology matching, ontology evolution and dynamics, modularization, etc.). Although an integral part of other research fields, such as text analysis, the full exploitation of ontologies for multimedia analysis is still largely unexplored.
The goal of this workshop is to gather three communities of researchers working in close, yet quite disconnected areas, related to semantic image interpretation: i) Image annotation and retrieval, ii) knowledge-based image analysis and iii) ontological engineering and reasoning including fundamental work on description logics or conceptual graphs.
| Submissions: | October 18, 2010 |
| Notification: | November 1, 2010 |
| Event: | December 1, 2010-December 3, 2010 |