The Use of Morphosemantic Regularities in the Medical Vocabulary for Automatic Lexical Coding
Susanne Wolff · Methods of Information in Medicine · 1984
Summary The Neo-Latin portion of medical vocabulary exhibits morphosemantic regularities that make it possible to determine both English syntactic subclasses and medical semantic subclasses from formal properties of these lexical items alone. This paper describes an experimental program for the automatic creation of dictionary entries that exploits the formal regularities to obtain dictionary entries sufficient for computerized medical text analysis as presently carried out by the New York University Linguistic String Project (LSP) system. Although automatic dictionary preparation does not supersede manual classification, the program takes a considerable burden off the dictionary worker’s shoulders and speeds the costly preprocessing stages of computerized text analysis.