Current scientific bibliographic databases (e.g., MEDLINE) now play a very central role in literature activities. They are important and they are widely used. However, the information expressed in the TITLE, ABSTRACT and MeSH INDEX fields in these databases is quite chaotic and sketchy compared with the structure of many other kinds of databases. Because of these deficiencies, computerized literature searches are imprecise and merely retrieve records based on the presence of """"""""keywords,"""""""" rather than via a precise """"""""match on concept"""""""" condition, such as is possible with relational databases and other orderly data structures. In this SBIR program we propose to make a small database containing the kind of information that is in ordinary abstracts, but to express it in a new """"""""scientific language"""""""" (MEDLISH), that we (ISG) have been developing. Unlike English, MEDLISH enforces order, scientific precision and logic on the entries. At the conclusion of Phase I we will have completed the design, made a sample database, and finished the implementation plans. In Phase II we then would make the marketable product. We expect that a data base expressed in MEDLISH will support much more precise queries than do current bibliographic databases, including, eventually, queries in natural English.
