[Elsnet-list] **Extended Deadline** ECAI-06 WS on Development and Evaluation of Spoken

ws-ecai06 at ling.uni-potsdam.de ws-ecai06 at ling.uni-potsdam.de
Tue Apr 18 18:31:55 CEST 2006

                    CALL FOR PAPERS

                      Workshop on


            in conjunction with ECAI 2006
       Riva del Garda, Italy, August 29th, 2006


The implementation of spoken dialogue systems, i.e. computer applications
capable to sustain a conversation with human users has long been a
challenging goal. Such man-machine interfaces need to apply knowledge from
all major fields of human language technology. The substantial
improvements in the area of speech recognition, dialogue modelling and
management, and speech synthesis have led to the recent development of a
range of deployed - and used - dialogue systems. The applications that
have been put forward mainly implement specific tasks ranging from voice
dialing and online plane ticket booking, to accessing information about
the weather and highway conditions, public transportation schedules and
local restaurants. These systems usually implement strictly
system-directed dialogues.

A few system architectures have been also proposed to sustain a less
restricted form of dialogue in narrow semantic domains. Such applications
are used, for instance, in the context of embodied conversational
characters for tutoring, educational or entertainment applications.
Despite the relative success of these spoken dialogue systems, much
research remains to be done on methods to create robust and feasible
mixed-initiative architectures that can also adapt or extend to new

The workshop will focus on the development and evaluation of robust spoken
dialogue systems, and in particular on the tension between the demands of
industrial applications on the one hand, and the creativity found in
laboratory research prototypes on the other. What can both sides learn
from one another? One of the important issues here is the search for
architectures capable to deal with understanding errors that may occur at
any stage of the processing pipeline between the speech recognizer and the
response generator.

The workshop intends to cover theoretical background from linguistics and
artificial intelligence on the nature of dialogue, as well as
interdisciplinary techniques and methodologies that are needed to cope
with the uncertainty in information processing. We aim to bring together
researchers and developers from different areas of natural language
processing and generation, leaning either to the industrial development or
to the reseach side, to discuss their experiences on the theoretical and
methodological aspects needed for robust dialogue interfaces and their
evaluation. Thus we solicit work from, but not limited to, the following

- Automatic Speech Recognition and Understanding
- Acoustic Plausibility
- Flexible (e.g. dynamic) grammars
- Dialogue Strategies
- Pragmatic Plausibility
- Error Detection and Recovery
- Strategies for Clarification Requests

Some of the questions we encourage contributors to address:

    * What role does or should dialogue modelling and management play in
the design of spoken dialogue systems?
    * How can a system detect its own errors?
    * What mechanisms should be implemented for recovering once errors
have been detected? What architectures are best appropriate for
implementing such mechanisms?
    * What are the experiences with respect to robustness and error
handling in deployed applications, and what are the ramifications for
research prototypes?
    * Is the appropriateness of the architecture a function of the purpose
of the dialogue system? If so, what associations exist between
dialogue models and dialogue systems?
    * What role does or could a spoken dialogue system play within AI
    * What learning strategy, if any, can be applied to the components of
a dialogue system to account for errors in other components?
    * What kind of strategy should be followed for system prompts and
clarification requests?

The workshop will take up one full day. It will be structured to allow for
maximum time for group discussion and participant interaction.




We welcome submissions for both oral (up to 8 pages) and poster
presentation (up to 4 pages). Papers must describe original, previously
unpublished research. Papers must be submitted electronically to
ws-ecai06 at ling.uni-potsdam.de in PDF format.


*NEW* 25th April, 2006
Workshop paper submission deadline.

10th May, 2006
Notification of paper acceptance.

18th May, 2006
Early registration deadline.

24th May, 2006
Paper camera-ready copy submission.

Tuesday 29th August, 2006
Workshop date.


Julie Baca, Mississippi State University, USA
Marc Cavazza, Teesside University, UK
Rachel Coulston, Oregon Graduate Institute for Science and Technology, USA
Daniele Falavigna, IRST, Italy
Roberto Gretter, IRST, Italy
Ed Kaiser, Natural Interaction LLC, USA
Sanjeev Kumar, Cisco Systems, USA
Lars Bo Larsen, Aalborg University, Denmark
Alpha Lee, Hong Kong Polytechnical University, Hong Kong
Boerge Lindberg, Aalborg University, Denmark
Carlos Martin-Vide, Rovira i Virgili University, Spain
Manish Mehta, Georgia Tech, USA
Sebastian Moeller, Deutsche Telekom Labs
Patrick Oliver, University of Newcastle upon Tyne, UK
Massimo Poesio, University of Essex
Markku Turunen, Tampere University, Finland
David Schlangen, University of Potsdam, Germany
Elena Zudilova, Amsterdam University, the Netherlands
Massimo Zancanaro, IRST, Italy


Andrea Corradini, University of Potsdam, Germany
Jörn Kreutel, Semantic Edge GmbH, Germany
Manfred Stede, University of Potsdam, Germany


Andrea Corradini (Workshop Chair)
Marie Curie Fellow (Tok-Dev Scheme)
University of Potsdam
Computational Linguistics Department
Karl-Liebknecht-Str. 24-25
14476 Golm
E-mail: ws-ecai06 at ling.uni-potsdam.de
URL: http://www.ling.uni-potsdam.de/ws-ecai06/

More information about the Elsnet-list mailing list