An associative-categorical model of word meaning

摘要

A new dual categorical-associative model for the representation of word meaning is proposed. In it, concepts are described by the values they have on a set of given variables (categories). A statistical relatedness measure (concomitant variation) is computed for these values on the basis of the specified word universe. An association measure between the word is defined, and the generalization of word clusters is introduced. A comparison with associative and categorical models is made and the application of the dual model to verbal analogy problems is described. Possible applications in Artificial Intelligence and Natural Language Processing are descussed.