The objective of this project is to develop the technology to produce compact and efficient expert systems. In support of this objective existing ai representation and processing techniques will be classified and optimization techniques for each one analyzed. The proposed Research will involve a survey of the literature, in addition to original Research. A software prototype will be constructed to demonstrate the feasibility of the concept. There are three primary benefits of this work. First, a compiler for an artificial intelligence (ai) language will be developed. Second, methods for generating finite state machines, which can be efficiencly implemented in terms of programmable read-only memories and microprocessors, for appropriate categories of expert systems will be developed. Third, the potential use of associative memories in conventional architectures will be determined. The commercial applications of this work are numerous. Such technology would allow the delivery of "real time" expert systems on cost-effective, general-purpose machines, with or without additional hardware devices.
