Bayesian reinforcement learning for POMDP-based dialogue systems | IEEE Conference Publication | IEEE Xplore