An experiment in computational discrimination of English word senses

Abstract

A number of researchers in text processing have independently observed that people can consistently determine in which of several given senses a word is being used in text, simply by examining the half dozen or so words just before and just after the word in focus. The question arises whether the same task can be accomplished by mechanical means. Experimental results are presented which suggest an affirmative answer to this query. Three separate methods of discriminating English word senses are compared information-theoretically. Findings include a strong indication of the power of domain-specific content analysis of text, as opposed to domain-general approaches.