8th International Conference on Spoken Language Processing

Jeju Island, Korea
October 4-8, 2004

A State Model for the Realization of Visual Perceptive Feedback in SmartKom

Peter Poller, Norbert Reithinger

DFKI, Germany

User acceptance of multimodal dialog systems depends on the quality of the process feedback to the user during interaction. We show an approach that permits the system to present simultaneously (and synchronously to the ongoing processing) adequate detailed and adaptive immediate feedback in various ways. For example, in multimodal dialog systems corresponding graphical presentations and/or appropriate behaviour patterns of an animated agent during user input processing significantly contribute to a more natural and expressively more powerful system. Additionally, the approach provides sufficient information to be able to present the user detailed descriptive feedback on the stage of processing at which an error occurred. The approach is mainly based on a module working state model that is used to determine in real time general processing state information in terms of relatively fine grained global system working states. Evaluations of the overall system show that simultaneous processing feedback is highly appreciated by users.

Full Paper

Bibliographic reference.  Poller, Peter / Reithinger, Norbert (2004): "A state model for the realization of visual perceptive feedback in smartkom", In INTERSPEECH-2004, 265-268.