The DIM system Plan Recognition in Spoken Dialogues

Umesh Berry, Washington University in St Louis
Michael Groner, Washington University in St Louis


Although speech recognition has improved significantly in recent years, it use in commercial environments has been limited by the inability of systems to model conventional co-operative dialogue. Any natural language systems for co-operative dialogue should at least be able to account for the many types of sub-dialogues that occur in conversations. Plan-based techniques provide a computationally powerful method for handling such dialogues. We discuss in this report our first attempt at building a plan-recognizer for an overall speech-driven system which interacts via a naturally restricted subset of English in a limited domain.