This research is primarily concerned with increasing the performance of expert systems. A refined focus of attention strategy and its affect on performance are discussed. Early expert systems used a brute force approach to process the knowledge base. Each production rule in the knowledge base was evaluated each cycle. More recently, processing efficiency has been increased by focusing the attention of the inference engine on a subset of the rules by "filtering" for further testing, only rules that could possibly fire given the current content of the context base. Focus of attention as developed in this research increases performance over filtering systems by further narrowing the focus of attention of the inference engine, down to the subexpression level. Positive results are reported.
Whiting, Kevin W.; DeKock, Arlan R.; and Prater, John Bruce, "A Focus Of Attention Algorithm For Expert Systems" (1984). Computer Science Technical Reports. 55.
© 1984 University of Missouri--Rolla, All rights reserved.
01 Jul 1984