Article Structure

Abstract

Open-ended spoken interactions are typically characterised by both structural complexity and high levels of uncertainty, making dialogue management in such settings a particularly challenging problem.

Introduction

The development of spoken dialogue systems for rich, open-ended interactions raises a number of challenges, one of which is dialogue management.

Background

2.1 Partially Observable Markov Decision Processes (POMDPs)

Approach

3.1 Control knowledge

Discussion

4.1 General comments

Conclusions

This paper described a new approach to exploit relational models of dialogue structure for controlling the action space in POMDPs.