In this talk I will first give an overview of the "Interpretation as Abduction" framework for interpreting natural language discourse. The interpretation of a broad range of linguistic phenomena is accounted for in this model in a thoroughly integrated fashion, including local pragmatic problems such as reference resolution and metonymy. The framework subsumes the "Parsing as Deduction" framework to cover syntactic processing. The recognition of discourse structure and reasoning about the speaker's plans can also be incorporated in a smooth fashion.

Abduction is implemented as a defeasible logic at the symbolic level. I will explore the possibilities and problems involved in implementing this logic in a connectionist architecture that uses temporal synchrony for dynamic variable binding, the SHRUTI architecture developed by Lokendra Shastri and his colleagues. I will then consider the problem of incrementally learning or modifying axioms, of the sort that would encode lexical, syntactic and commonsense knowledge, at both the symbolic and the connectionist levels.

The enterprise outlined here looks very promising, and if successful, it would give us at least a plausible sketch of the link between an interesting class of intelligent behavior and neurophysiological activity.

Bio:

Dr. Jerry R. Hobbs is a prominent researcher in the fields of computational linguistics, information extraction, discourse analysis, and artificial intelligence. He earned his doctor's degree in computer science from New York University in 1974. He has taught at Yale University and the City University of New York. Since 1977 he has been with the Artificial Intelligence Center at SRI International, Menlo Park, California, where he is now a principal scientist and the program director of the Natural Language Program. He is the author of numerous papers and of the book "Literature and Cognition", and has also edited the book "Formal Theories of the Commonsense World". He is a consulting professor with the Linguistics Department at Stanford University. He has served as general editor of the Ablex Series on Artificial Intelligence. He is a past president of the Association for Computational Linguistics, and has been elected a Fellow of the American Association for Artificial Intelligence.