Natural Language Processing
University of Melbourne, 10-14 July 2006
|
|
Natural Language Processing University of Melbourne, 10-14 July 2006 |
|
Although it is now possible to build real-time spoken dialogue systems for a wide variety of applications, designing the dialogue strategy of such systems involves a number of nontrivial design choices. These choices can seriously impact system performance, and include choosing appropriate strategies for initiative (whether to accept relatively open-ended vs. constrained user utterances) and confirmation (when to confirm a user's previous utterance). Typically, dialogue strategy design has been done in an ad-hoc manner, with subsequent improvements to dialogue strategy being fielded sequentially.
In this tutorial, we will address the design, implementation, and experimental evaluation of methods for learning and optimizing a system's dialogue strategy, in order to improve system performance. We will show how to apply the formalism of Markov decision processes (MDPs) and the algorithms of reinforcement learning to the problem of automated dialogue strategy synthesis. In this approach, an MDP is built from training data gathered from an initial "exploratory" system. This MDP provides a state-based statistical model of user reactions to system actions, and is used to simultaneously evaluate many dialogue strategies and choose the apparent optimal among them. We will discuss how to run controlled user experiments to evaluate the approach. We will conclude with advanced topics in this area, including user simulation and issues in state representation.
Diane Litman is Professor of Computer Science, as well as Research Scientist with the Learning Research and Development Center, at the University of Pittsburgh. Previously, Dr. Litman was a member ofthe Artificial Intelligence Principles Research Department, AT&T Labs - Research (formerly Bell Laboratories); she was also an Assistant Professor of Computer Science at Columbia University. Dr. Litman received her Ph.D. degree in Computer Science from the University of Rochester. Her current research focuses on enhancing the effectiveness of tutorial dialogue systems through the use of spoken language processing, affective computing, and machine learning. From 2000-2003 she was Chair of the North American Chapter of the Association for Computational Linguistics.