Dialogue Management for a Multi-Party Spoken Dialogue System



This chapter is dedicated to the multi-party dialogue management functionality of our dialogue system which denotes the most important part of this book. The dialogue management deployed is built on the basis of the Information State Update approach (e.g. [Cooper and Larsson, 1998,Traum et al., 1999, Matheson et al., 2000, Larsson and Traum, 2000, Larsson, 2002]), as introduced in Section 2.4. The ISU approach models dialogue as a state of information that is updated according to the content of the latest incoming utterance. It is very well suited to model an agent-like system as ours which is to stand independently as a conversation partner. The system's (and also the other users') state of 'mind' including believes about itself and the others as well as the goals it aims to achieve can be modelled. The approach provides exibility for the developer to decide what information should be specified and in what way. It allows modifications and adjustments and thus can be adopted to fit the requirements of our proactive multi-party dialogue system. Modi_cations of the ISU approach become necessary due to the fact that the existing multi-party extensions to the approach do not fully meet the requirements of our system, as described in Section 2.4.4.


Task Model Database Query Dialogue System Current Preference Dialogue Management 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag US 2010

Authors and Affiliations

  1. 1.Institute of Information TechnologyUlm UniversityUlmGermany

Personalised recommendations