Simulating Rule-Based Systems
AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OH SCHOOL OF ENGINEERING
Pagination or Media Count:
The purpose of this thesis is to develop a methodology for evaluating the performance of rule-based systems RBSs using a simulation approach. A numerical scheme is used for knowledge representation facts are represented by integer numbers and the rules and data memories are represented by matrices. The numeric representation can be handled by simplified algorithms that simulate the function of different types of inference engines. Six types of forward-chaining inference engines that vary according to the conflict resolution strategy and the implementation of filters are simulated and compared. The number of match- tests of the left-hand side of the rules against the data memory is used as a measure of performance to estimate the relative matching effort for each inference engine. Also, a methodology to reduce the matching effort of RBS by changing the order of the facts in the left-hand side or changing the order of the rules is described. Keywords Expert systems, Computer programs.
- Operations Research