The major objective of the project is the development of methods for the automatic processing of natural medical language. Research in medical linguistics includes the development of paraphrasing rules for medical noun phrases, a semantic grammar, and a generalized morphological analysis approach suitable for general language and data compression as well as medical language morphology. Development of the Lexicographic Environment Software, a platform for medical lexicography and medical language processing, has continued. Work has begun on a study of error backpropagation in neural networks to determine what changes would be necessary to allow facile solution of harder problems such as ambiguity of input symbols (e.g. the letter """"""""c"""""""" pronounced in hard and soft variants) and the exclusive-or problem. Collaboration continued on the Clinical Information Utility with the Laboratory of Pathology (NCI/DCBD/LP), and with DCRT/DMB, in using LSM's automatic encoding system to maintain and improve the data base of Clinical Center surgical pathology reports. This year, it has been possible to update the encoded data base and do medical lexicography with the LP staff on a weekly basis. Efforts were made with D.E. Henson to identify linguistic regularities in 296 pathology reports by 8 dertmatopathologists engaged in a revision of the terminology for borderline malignant melanoma at the request of A. Moshell. Collaboration has begun with G. Michaels concerning the rapid prototyping of molecular biology research aids. A translator (in the C language) for Genbank data to the Prolog language has been written. Other linguistic expertise may prove valuable in genetic sequence analysis.
