The Representation and Selection of Commonsense Knowledge for Natural Language Comprehension.

摘要

Representation and selection of commonsense knowledge about cause and effect are central aspects of human intelligence. This report describes a representation technique called Commonsense Algorithms, and describes the organization of knowledge encoded in this scheme. Current research is focussed on three major areas: the application of the representation formalism to (1) the comprehension of a children's story called The Magic Grinder: (2) the representation and simulation of devices and mechanisms, and (3) problem solving. Selectional issues are discussed with respect both to the commonsense algorithm representation, and to word sense selection in natural language parsing. (Author)