A Representation for Natural Category Systems

Abstract

Most AI systems model and represent natural concepts and categories using uniform taxonomies, in which no level in the taxonomy is distinguished. We present a representation of natural taxonomies based on the theory that human category systems are non-uniform. There is a basic level which forms the core of a taxonomy; both higher and lower levels of abstraction are less important and less useful. Empirical evidence for this theory is discussed, as are the linguistic and processing implications of this theory for an artificial intelligence/natural language processing system. Among these implications are: (1) when there is no context effect, basic level names should be used; (2) systems should identify objects as members of their basic level categories more rapidly than as members of their superordinate or subordinate categories. We present our implemen-tation of this theory in SNePS. a semantic network processing system which includes an ATN parser-generator, demonstrating how this design allows our system to model human performance in the natural language generation of the most appropriate category name for an object. The ability of our system to acquire classificational information from natural language sentences is also demonstrated.