An automated indexing system utilizing semantic net expansion

Automated subject indexing and the coordination of indexing with natural language search interfaces could provide solutions to some of the information retrieval problems facing the medical field today. This paper describes the construction of such an interface using semantic net expansion and a clinical neuroscience thesaurus, which provided a specialized vocabulary for use in both automated indexing of MEDLINE articles and an "intelligent" front-end for database searching. Though reviewed by a domain expert, the thesaurus was successfully built by nonexperts utilizing a set of rules. We report on testing thesaurus content and semantic net accuracy using a database of selected MEDLINE article abstracts.