An integrated natural language processing (NLP) and information retrieval (IR) system will be prototyped for text retrieval from medical text databases or large document browsing. The integrated NLP/IR system will utilize a parser with word sense disambiguation, morphological analysis, an inverted """"""""word sense"""""""" index over the documents, and other advanced features identified by recent information retrieval workshops. Several machine readable dictionaries will be integrated, including Miriam Webster, Princeton's WordNet, and the UMLS Meta-thesaurus and mesh terms from NLM. This integrated resource will be used as a knowledge base for processing natural language information retrieval requests for medical texts. Retrieval experiments will be conducted to estimate the precision and recall of various methods of query. This project has significant commercial application and it makes a potential contribution to the UMLS. Experiments will be conducted on the MEDLINE database.
