| Rule-based systems represent one of the most popular paradigms in the artificial intelligence area. However, the issues of completeness and consistency in rule-based systems have been largely ignored in the development of expert systems. Some research has been carried out to solve the problems of consistency and completeness. But most of the current methods use tables or enumeration only methods to check a rule set for consistency and completeness.;This research suggests a new Tree-Based approach to overcome some of the difficulties of the current rule verification methodology by avoiding enumeration only methods and rule-to-rule comparison in testing for completeness and consistency. This research uses a K-tree representation to test the completeness and consistency of a rule base.;Once a set of rules is represented as a tree, several search techniques such as breadth-first search and depth-first search can be used to detect missing rules, redundant rules, conflict rules, and subsuming rules. Pruning branches in a tree during the search process significantly reduces the search space and, thereby, makes testing for completeness and consistency far more efficient than other methods. The detailed algorithm of search to test for completeness and consistency is investigated in this research.;The main advantage of the suggested method is that it avoids a rule-to-rule comparison, which often causes combinatorial explosion problems when there are many parameters and possible values involved. It also avoids enumeration only methods which display all the possible combinations of values of parameters. This new approach can be applied to most rule-based systems or rule-based system shells because the design is general. Adding such an additional rule base checking capability to rule-based systems would significantly increase the accuracy and the performance of rule-based systems, which would be the main contribution of this research. |
