Modeling Time and Meta-Reasoning in Dialogue Via Active Logic

Contents

We describe our work on human-computer cooperative dialogue, that uses special "active" logics in an attempt to more closely - and more usefully - model human behavior. This effort pays attention to a number of issues, including Gricean pragmatics, logical nonomniscience, reasoning with inconsistent data, limited memory, blending in new information during reasoning, and models of human communication competence. Time is a chief unifying feature in our approach to these issues.