Probabilistic methods in spoken-dialogue systems

This paper presents a probabilistic framework for modelling spoken-dialogue systems. On the assumption that the overall system behaviour can be repres ented as a Markov decision process, the optimization of dialogue-management strategy using reinforcement learning is reviewed. Examples of learning be haviour are presented for both dynamic programming and sampling methods, bu t the latter are preferred. The paper concludes by noting the importance of user simulation models for the practical application of these techniques a nd the need for developing methods of mapping system features in order to a chieve sufficiently compact state spaces.