Word formation in natural language processing systems
Roy J. Byrd · International Joint Conference on Artificial Intelligence · 1983
Systems which process natural language require a reliable source of information about words. Not only must their lexical subsystems handle a large number of known words; they must also cope with coinages. The morphological principles underlying the notion possible are under active study by linguists, and are articulated in the theory of word formation. This paper presents a technique for building lexical subsystems which embody these principles by emulating the behavior of word formation rules. These subsystems combine totally idiosyncratic lexical information, stored in a dictionary, with systematic information derived from word structure. Applications for lexical subsystems built along the lines described here will be discussed.