User simulations for online adaptation and knowledge-alignment inTroubleshooting dialogue systems Export

@inproceedings{citeulike:4882876, abstract = {We study the problem of alignment between dialogue participants, using the practical example of ” troubleshooting” dialogue systems. Recent work on troubleshooting concerns automated spoken dialogue systems which support users who need to repair their internet connection. We address the problem that different users have different types of knowledge of problem domains, so that automated dialogue systems need to adapt online to the different knowledge of these users as it encounters them. We approach this problem using policy learning in a Markov Decision Process (MDP). In contrast to related work we propose a new user model which incorporates the different conceptual knowledge of different users, together with an environment simulation. We show that this model allows us to learn dialogue policies that automatically adapt online to new users, and that these policies are significantly better than threshold-based adaptive hand-coded policies for this problem.}, author = {Janarthanam, Srinivasan and Lemon, Oliver}, booktitle = {Proceedings of LONDIAL: The 12th SEMDIAL Workshop}, citeulike-article-id = {4882876}, keywords = {dialogue-management, knowledge-alignment, reinforcement-learning}, posted-at = {2009-06-17 18:57:00}, priority = {2}, title = {User simulations for online adaptation and knowledge-alignment inTroubleshooting dialogue systems}, year = {2008} }

TY - CONF ID - citeulike:4882876 L3 - citeulike-article-id:4882876 N2 - We study the problem of alignment between dialogue participants, using the practical example of “troubleshooting” dialogue systems. Recent work on troubleshooting concerns automated spoken dialogue systems which support users who need to repair their internet connection. We address the problem that different users have different types of knowledge of problem domains, so that automated dialogue systems need to adapt online to the different knowledge of these users as it encounters them. We approach this problem using policy learning in a Markov Decision Process (MDP). In contrast to related work we propose a new user model which incorporates the different conceptual knowledge of different users, together with an environment simulation. We show that this model allows us to learn dialogue policies that automatically adapt online to new users, and that these policies are significantly better than threshold-based adaptive hand-coded policies for this problem. TI - User simulations for online adaptation and knowledge-alignment inTroubleshooting dialogue systems T2 - Proceedings of LONDIAL: The 12th SEMDIAL Workshop KW - dialogue-management KW - knowledge-alignment KW - reinforcement-learning AU - Janarthanam, Srinivasan AU - Lemon, Oliver PY - 2008/// ER -