Software Integration of Real-Time Expert Systems

Abstract A technique to achieve real-time computing throughput from rule-based expert systems is discussed. Using compilation, compaction, and parallelization techniques, this research synthesizes a set of concurrently executable Ada tasks from a knowledge’ base of rules. A prototype compilation system based on this approach has demonstrated speedups in excess of 100X along with increased embeddability of the knowledge base. The work is being extended into a software engineering development environment for expert systems whereby programming constructs from both the procedural and rule-based language domains are made available to the user. The approach strives to present the engineer with programming templates that protect him from the intricate control mechanisms of the inference engine and to enable him to concentrate on problem solving at higher levels of abstraction.