SPONSORS:
National Science Foundation
PROGRAM COMMITTEE:
Alexander I. Rudnicky, Carnegie Mellon University, Co-chair
Candace L. Sidner, Mitsubishi Electric Research Laboratories, Co-chair
CONFERENCE WEBSITE:
http://www.speech.cs.cmu.edu/hltnaacl2003/
PREFACE
The purpose of this workshop is to provide a forum for discussion of current directions in dialog
research, specifically to assess the current state of the art in the area of dialog processing, and to identify
key themes and directions that are driving research in the field. The motivation to do so at this time
is the need to establish the role of dialog as a core element in human-human and human-computer
communication and to define its role in the forthcoming NSF Human Language Communication program.
At this time, many researchers and product engineers are beginning to develop dialogue capabilities
for spoken language systems. Their experience is rapidly creating information and artifacts which in
turn are attracting increasing interest on the part of researchers from a variety of disciplines. One reason
for this ferment is that two groups who traditionally have had little to do with each other, linguists /
computational linguists and engineers, each approaching dialog from different perspectives, have begun
to interact on a technical level. In part this is due to the emergence of working technologies, such as
recognition systems and speech synthesizers, that for the first time allow researchers not directly familiar
with the implementation of component technologies to put together systems that converse (however
simply) with humans. As a result, groups with very different traditions now find themselves work-
ing on phenomena that are nominally the same. These researchers are concerned about making use
of linguistically motivated dialogue models, the need for well-engineered, practical interfaces for use
with everyday users, and the availability of corpora that can steer new research in this area for both
computational linguists and engineers. These shared concerns present an opportunity to encourage cross-
fertilization and to transform the study of dialog into a richer and more energetic enterprise. In turn, such
a transformation will increase our understanding of dialog and will hasten the creation of techniques and
artifacts that significantly impact human-computer communication.
We would like to thank the authors and the members of the Program Committee for their contributions
to the planning and execution of the workshop, the workshop sponsor, the National Science Foundation
for their financial support, in particular Mary Harper and Karen Kukich, as well as the HLT/NAACL
conference organizers, especially Ed Hovy, James Allen, Steven Abney, and Dragomir Radev, for their
significant contributions to the overall management of the workshop and their direction in preparing the
publication of the proceedings.
Alex Rudnicky and Candy Sidner May 2003
