Online planning algorithms for POMDPs
Journal of Artificial Intelligence Research
Planning and acting in partially observable stochastic domains
Artificial Intelligence
Hi-index | 0.00 |
In this paper we will describe the generic module of a dialogue manager that we call the BigList. It is intended to solve the isolated task of selecting one item from the huge list of items with assigned attributes. The module incorporates the asking strategy based on partially observed Markov decision process. The model uses some assumptions and simplifications to achieve the computational tractability. The module is intended to completely generalize the common task of selecting the concrete item from database in spoken dialogue systems.