Special Sessions

There will be one special session co-located with the 8th International Conference on Natural Language Generation (INLG 2014) on the afternoon of June 19. There will be a second special session on the Dialog State Tracking Challenge (DSTC) on the morning of June 20. Special session submissions will undergo regular SIGDIAL review process.


Submission of special session proposals is now closed.

The SIGDIAL organizers welcome the submission of special session proposals. A SIGDIAL special session is the length of a regular session at the conference; may be organized as a poster session, a poster session with panel discussion, or an oral presentation session; and will be held on the last day of the conference. Special sessions may, at the discretion of the SIGDIAL organizers, be held as parallel sessions. Those wishing to organize a special session should prepare a two-page proposal containing: a summary of the topic of the special session; a list of organizers and sponsors; a list of people who may submit and participate; and a requested format (poster/panel/oral session). These proposals should be sent to conference[at]sigdial.org by the special session proposal deadline. Special session proposals will be reviewed jointly by the general and program co-chairs.


The Dialog State Tracking Challenge (DSTC) is a research challenge focused on improving the state of the art in tracking the state of spoken dialog systems. State tracking, sometimes called belief tracking, refers to accurately estimating the user's goal as a dialog progresses. Accurate state tracking is desirable because it provides robustness to errors in speech recognition, and helps reduce ambiguity inherent in language within a temporal process like dialog.

This special session covers results from the second instance of this challenge (the first instance was covered last year at SIGDIAL 2013). The dialogs for this challenge were in the restaurants domain and were collected with Amazon Mechanical Turk. Participants were given labelled corpora of dialogs to develop state tracking algorithms. The trackers were then evaluated on a common set of held-out dialogs which were released, un-labelled, during a one week period. Nine teams participated in the challenge. This special session will cover results from this challenge task.

More information is available at the DSTC website: http://camdial.org/~mh521/dstc/.

Matthew Henderson, University of Cambridge, UK
Blaise Thomson, University of Cambridge, UK
Jason Williams, Microsoft Research, USA

Last update: May 8, 2014 - Credits