NAACL HLT 2012

Workshop on Future directions and needs in the Spoken Dialog Community: Tools and Data

June 7, 2012, Montreal

Working Notes (including position statements and project notes)

Schedule

Description of the Workshop

The goal of this workshop is to determine what the present and future common needs of the spoken dialog research community are.

The spoken dialog research community has interests in a wide variety of topics such as belief tracking, multimodal dialog, simulated users, dialog assessment, etc. This diversity of interests makes it difficult to find common threads that can bind the community together. We can cite three types of existing common tools and data. One is data sharing. Another is the Spoken Dialog Challenge, which enables many researchers to compare techniques on a common task. Yet another is the use of a well-documented platform that can be used to teach new researchers about spoken dialog architecture, thus enabling easier entry to our field, a project which has also begun. There may be other needs that have emerged in the community. This workshop will describe the present state of tools and data, and explore upcoming needs.

This workshop will be sponsored by the Dialog Research Center (DialRC), a project funded by the NSF Community Research Initiative (CRI) Program, which has served the community by providing data, a teaching tool platform and a common challenge task.

The workshop encourages papers on the state of the art in spoken dialog and needs in data and tools. The areas suggested are:

  • dialog system architecture (new architectures, modules, etc.)
  • speech recognition for spoken dialog systems (noise robust telephone ASR, etc)
  • spoken language understanding
  • dialog management (plan-based, information state update, turn-taking, error handling, grounding, etc)
  • statistical machine learning (reinforcement learning, user simulation, belief tracking, etc)
  • speech synthesis (conversational articulatory synthesis, natural language generation, etc)
  • tutoring systems (ESL task-oriented systems, etc)
  • human-robot dialog, dialog systems in 3D virtual worlds
  • dialog system assessment
and other related areas

In the morning, speakers will cover the main advances that have been made in that area in the past five years and then point to specific promising topics of research. Noted spoken dialog researchers will then organize breakout groups on these topics and the groups will report back to all of the participants to attempt to define needs that are common across topics and thus the most widespread in the community.

Important Dates

Initial Submission 11:59 PDT April 8th, 2012
Notification: April 24th, 2012
Final submission: May 7, 2012
Workshop June 7, 2012

Submissions

Potential participants should submit one or more of the following:
  • a short project note on existing tools (no more than 4 pages, for possible presentation during the poster session)
  • position statements on either future directions and needs and/or ideas for future tools or efforts to address these needs (no more than two pages, to be posted in advance, possibly summarized in the morning talks, and discussed in the breakout groups)

    All submissions should be in NAACL format (but submissions need not be anonymous).

    Submit papers to https://www.softconf.com/naaclhlt2012/SDCTD2012/

    Organizing Committee

    • Maxine Eskenazi Carnegie Mellon
    • Alan Black Carnegie Mellon
    • David Traum USC Institute for Creative Technologies

    Scientific Committee

    • Dan Bohus Microsoft
    • Joyce Chai Michigan State University
    • Dilek Hakkani-Tur Microsoft
    • Helen Hastie Heriot-Watt University
    • Julia Hirschberg Columbia
    • Kristiina Jokinen University of Helsinki
    • Diane Litman University of Pittsburgh
    • Helen Meng Chinese University of Hong Kong
    • Wolfgang Minker University of Ulm
    • Sebastien Moeller Deutsche Telecom
    • Olivier Pietequin Supelec
    • Antoine Raux Honda
    • Giuseppe Riccardi University of Trento
    • Amanda Stent AT&T
    • David Suendermann SpeechCycle
    • Nigel Ward University of Texas El Paso
    • Jason Williams AT&T
    • Steve Young Cambridge University