ARTIFICIAL-INTELLIGENCE LOOKS AT NATURAL-LANGUAGE

For researchers in artificial intelligence, a central problem in processing natural languages is that of building an internal representation of the meanings of discourses. This process of semantic interpretation involves grammar-based and knowledge-based processes. Grammar-based processes perform lexical access and syntactic analysis. Knowledge-based processes are necessary for resolving ambiguities, for determining specific referents for noun phrases, for identifying extensions of meanings, and for developing a globally coherent model of the situation described in the discourse. Knowledge-based processes very much depend on being able to make the kinds of common sense inferences that humans make to fill in what is not explicit in the discourse. The way the internal representation of discourse content is organized is at the heart of the design of any system that can understand natural language discourse. © 1990.