Versatile RDF Representation for Multimedia Semantic Search

The integration of semantic representation and retrieval technologies into mainstream Web applications depends on the ease of adoption and re-use of existing information and meta-data. While for textual information, analysis of the text content is quite standardized, for multimedia resources many archives and repositories resorted do define different metadata representations, creating obstacles to interoperability. The flexibility of RDF allows representing and managing all kind of meta-data and of semantic information about multimedia resources, and this paper proposes a possible strategy, based on a structure called RDF Descriptor, that allows representing, reconciling and semantically tagging multimedia resources of different media formats, and possibly coming from different sources with different representations. Experimental results show the feasibility of the approach by reporting result on cross-media and cross-archive semantic searches.