AN INFORMATIONAL NETWORK FOR A NATURAL TALKING SYSTEM

A Natural Language Processing System has been developed that communicates in the simple commonsense terms of a child. The system utilizes a knowledge representation scheme that facilitates the storage and retrieval of factual and inferential data through a massively-interlinked network of propositional information. Inferential analysis is handled by an organized matrix of predicate-based inferences that uses forward chaining to deduce new propositions. The potential for learning has been established through inductive generalization and the ongoing adjustment of uncertainty values on existing propositions.