Computational pragmatics for face-to-face dialogue
Lecturer(s):Matthew Stone (Rutgers University) and David Traum (Institute for Creative Technology, USC) and Justine Cassell (Northwestern University)
Type:Advanced Course
Section:Language and Computation
Time: 9.00-10.30 (Slot 1)
Room:EM 1.27


Interactive dialogue systems increasingly pursue situated
communication about ongoing events, and present interlocutors with an
embodied partner, perhaps a graphical character in a virtual world, or
perhaps a physical robot.  For such conversational systems, it is
particularly important to work actively with interlocutors to maintain
an evolving shared context.  This course offers an overview of current
computational models (including publicly-available software tools and
platforms) that account for the embodied communicative actions such
systems can recognize and use.  We will look at a range of behaviors,
including gaze and posture, gesture, and demonstration, as well as
spoken utterances.  We will show how these behaviors can serve in
implemented systems to achieve a range of context-management
functions: establishing presence and shared attention to a
conversation; coordinating speaking turns; presenting and grounding
individual contributions to conversation; handling multiple
conversations and other concurrent events; and achieving coordinated
goals of information exchange, decision making, and action.

© ESSLLI 2005 Organising Committee 2004-12-01