Lexikos has prototype common lisp software able to systematically find, model, and relate the central topics of well-formed English text, then record this data in a modular way so that other software can access it easily. We propose using an enhanced version of this code to show the: feasibility of a high-volume production of expert systems such as those cited in the SBIR solicitation. Instead of using knowledge engineers ( a scarce resource) to create them, our semi-automated methods would substitute existing text as the information resource and knowledgeable clerical staff as its editors. Each new expert system would be initiated by having our final-version software scan and analyze one or more existing English documents __ training manuals, system specs, etc. From them, it would "extract" the vocabulary, definitions, relationships, and facts bounding the intended problem domain. The results of this brief process, combined with presentation software and mission-dependent extensions, would become the first version of the new expert system's knowledge base, and provide a solid foundation for expansion.
