Retrieval of VidEo And Language for The Home user in an Information Society

REVEAL THIS

Retrieval of VidEo And Language for The Home user in an Information Society
The REVEAL THIS project aimed at doing research on and developing a multimedia and multilingual content processing system, i.e. a system that would process TV and radio programmes and web documents in order to index them as effectively as possible for easy access to this content by all. The prototype that has been developed is able to handle both English and Greek documents pertaining to two domains: politics (mainly European Parliament sessions, corresponding press conferences and national news) and travel (mainly documentaries and informative travel documents on various holiday destinations). The prototype allows for cross-lingual information retrieval through a web interface and for content filtering for pushing information to a mobile phone user according to his/her preferences. The user may watch only that part of a programme that is related to his/her query, or even the whole programme if the multimedia presentation/summary (s)he is presented with attracts his/her attention. The REVEAL THIS prototype combines medium-specific processing modules (i.e. automatic speech recognition, named entity recognition, term and fact extraction, face detection and identification, speaker identification, image and text categorization, machine translation) for performing cross-media indexing-retrieval-filtering and summarization of multimedia documents.
Status
Completed
Start Date
End Date
Partners

BeTV (Belgium)
Katholieke Universiteit Leuven (Belgium)
Sail Labs Technology AG (Austria)
TVEyes (USA)
University of Strathclyde (United Kingdom)
Xerox Research Centre Europe (France)
Ινστιτούτο Επεξεργασίας του Λόγου / Ε.Κ. "Αθηνά"