Semantic modelling using TV-anytime genre metadata

Abstract

The large amounts of TV, radio, games, music tracks or other IP based content becoming available in DVB-H mobile digital broadcast, offering more than 50 channels when adapted to the screen size of a handheld device, requires that the selection of media can be personalized according to user preferences. This paper presents an approach to model user preferences that could be used as a fundament for filtering content listed in the ESG electronic service guide, based on the TVA TV-Anytime metadata associated with the consumed content. The semantic modeling capabilities are assessed based on examples of BBC program listings using TVA classification schema vocabularies. Similarites between programs are identified using attributes from different knowledge domains, and the potential for increasing similarity knowledge through second level associations between terms belonging to separate TVA domain-specific vocabularies is demonstrated. © Springer-Verlag Berlin Heidelberg 2007.