Plan Recognition and Its Use in Understanding Dialog

Abstract

Analysis of naturally occurring dialog suggests that human information-providers often use knowledge acquired from the dialog to understand subsequent utterances and provide cooperative, helpful responses. One of the most important ways in which an information-seeking dialog can be assimilated is by inferring the task-related plan motivating an information-seeker’s queries. This chapter presents two process models. The first dynamically constructs a model of an information-seeker’s plans and goals from an ongoing dialog, and the second reasons on this component of a user model to understand one class of problematic utterance. The chapter further addresses the problem of detecting and recovering from discrepancies between the system’s model of the user and the actual plan under construction by the user. It proposes a four-phase approach to detecting and recovering from disparate models, and argues for an enriched user model that differentiates among its components on the basis of the support it accords each component as a correct and intended part of the user’s plan.