9th Annual Conference of the International Speech Communication Association

Brisbane, Australia
September 22-26, 2008

Effects of User Modeling on POMDP-Based Dialogue Systems

Dongho Kim (1), Hyeong Seop Sim (1), Kee-Eung Kim (1), Jin Hyung Kim (1), Hyunjeong Kim (2), Joo Won Sung (2)

(1) KAIST, Korea; (2) KT, Korea

Partially observable Markov decision processes (POMDPs) have gained significant interest in research on spoken dialogue systems, due to among many benefits its ability to naturally model the dialogue strategy selection problem under the unreliability in automated speech recognition. However, the POMDP approaches are essentially model-based, and as a result, the dialogue strategy computed from POMDP is subject to the correctness of the model. In this paper, we extend some of the previous user models for POMDPs, and evaluate the effects of user models on the dialogue strategy computed from POMDP.

Full Paper

Bibliographic reference.  Kim, Dongho / Sim, Hyeong Seop / Kim, Kee-Eung / Kim, Jin Hyung / Kim, Hyunjeong / Sung, Joo Won (2008): "Effects of user modeling on POMDP-based dialogue systems", In INTERSPEECH-2008, 1169-1172.