Thursday, November 12, 1998, Volen 101, 2:00-3:00pm
n this talk, I describe an experimental method by which spoken dialogue agents can learn to choose an optimal dialogue strategy from interacting with human users. Previous work has suggested a number of heuristics that can guide an agent's selection of dialogue strategies when there are multiple ways to realize a communicative intention, but there has been little work on automatically optimizing an agent's choices. Our method is based on a combination of learning algorithms and empirical evaluation techniques. The learning component of our method is based on algorithms for reinforcement learning, such as dynamic programming and Q-learning. The empirical component uses the PARADISE evaluation framework (Walker et al., 1997) to identify the important performance factors and to provide the performance function needed by the learning algorithm. We illustrate our method with a dialogue agent named ELVIS (EmaiL Voice Interactive System), that supports access to email over the phone. We show how ELVIS can learn to choose among alternate strategies for agent initiative, for reading messages, and for summarizing email folders.
Host: Rick Alterman