Title :
Spoken dialogue management as planning and acting under uncertainty
Author :
Zhu, Xin-zhong ; Zhao, Jian-Min
Author_Institution :
Inst. of Comput. Sci. Studies, Zhejiang Normal Univ., China
Abstract :
Some stochastic models like the Markov decision process (MDP) are used to model the dialogue manager. MDP-based systems degrade faster when uncertainty about the user´s intention increases. We propose a novel dialogue model based on the partially observable Markov decision process (POMDP). We use hidden system states and user intentions as the state set, parser results and low-level information as the observation set, and domain actions and dialogue repair actions as the action set. Low-level information is extracted from different input models using Bayesian networks. Because of the limitations of exact algorithms, we focus on heuristic methods and their applicability in dialogue management.
Keywords :
Markov processes; belief networks; decision theory; speech processing; uncertainty handling; Bayesian networks; Markov decision process; acting; action set; dialogue repair actions; domain actions; heuristic methods; hidden system states; low-level information; observation set; parser results; partially observable Markov decision process; planning; spoken dialogue management; state set; stochastic models; uncertainty; user intentions; Bayesian methods; Bridges; Computer science; Electronic mail; Knowledge management; Probability distribution; Process planning; Resource management; Stochastic processes; Uncertainty;
Conference_Titel :
Machine Learning and Cybernetics, 2002. Proceedings. 2002 International Conference on
Print_ISBN :
0-7803-7508-4
DOI :
10.1109/ICMLC.2002.1174420