Homepage IMIX project VIDIAM,
dialogue management and the visual channel
LATEST UPDATE 27 feb 2008
On this page you will find the newest information about VIDIAM.
IMIX is a multimodal dialogue system for question answering. IMIX answers
factual questions in a medical domain, generating its answers in multimedia
(speech, text, pictures, videos). The user may use speech, text, and pen
strokes on a tablet. It is a collaboration of 7 research projects. VIDIAM
concerns the dialogue manager.
VIDIAM collaborates with PARADIME and ROLAQUAD to form a dialogue manager.
VIDIAM emphasises multimodality and wishes to combine state of the art
techniques from multimodal dialogue systems and qa dialogue systems. PARADIME
emphasises the possibilities of direct implementation of Dynamic
Interpretation Theory using a production rule type system. ROLAQUAD wishes to
deliver classifier modules for the dialogue manager, such as dialogue act
classification and problematic communication indication.
People
Rieks op den Akker, coordinator
Boris van Schooten, postdoc
Hilco Kats has done a D assignment on
dialogue managers for clarification dialogues.
(homepage)
Boris van Schooten and Rieks op den Akker, Follow-up utterances in QA
dialogue. In: Traitement Automatique des Langues, 46(3), 2005
(Appeared in January 2007). Pages 181-206.
Mariet Theune, Emiel Krahmer, Boris van Schooten, Rieks op den Akker,
Charlotte van Hooijdonk, Erwin Marsi, Wauter Bosma, Dennis Hofs and Anton
Nijholt.
Questions, Pictures, Answers: Introducing Pictures in Question-Answering
Systems
In: Tenth international symposium on
social communication, Cuba, January 2007. Pages 450-463.
Rieks op den Akker and Boris van Schooten,
Vervolguitingen in vraag-antwoorddialoogsystemen.
In DIXIT, tijdschrift over toegepaste taal- en spraaktechnologie, 4(1), pages 10-11, June 2006.
QAC2 Task3 dialogues: (see also the
overview page). Japanese | Automatic English translation. Note: the
1st and 3rd columns contain the actual questions, the 2nd column contains the
self-contained version of the questions.
SmartKom multimodal
corpus. This is a video corpus. Transcriptions of the dialogues are
downloadable though.