A Belief Tracking Challenge Task For Spoken Dialog Systems
- Jason Williams
NAACL HLT 2012 Workshop on Future directions and needs in the Spoken Dialog Community: Tools and Data |
Published by Association for Computational Linguistics
Belief tracking is a promising technique for adding robustness to spoken dialog systems, but current research is fractured across different teams, techniques, and domains. This paper amplifies past informal discussions (Raux, 2011) to call for a belief tracking challenge task, based on the Spoken dialog challenge corpus (Black et al., 2011). Benefits, limitations, evaluation design issues, and next steps are presented.