Performs implication using correlation rules.
Namespace: FuzzyInterfaces
Delphi |
public |
Parameters
Premise
Type: IFuzzySet
NewEvidence
Type: IFuzzySet
Matching
Type: IFuzzySet
OldEvidence
Type: IFuzzySet
Type: IFuzzySet
With correlation implication rules, the consequent fuzzy set is correlated with the premise fuzzy set's truth value, which is modified by the Matching fuzzy set. Basically, with the AssertRule method, you have a fuzzy fact (the OldEvidence parameter). The rule is formed as "If PREMISE then NEWEVIDENCE". Then, you give the AssertRule method some fact which matches the premise (the Matching parameter) and it creates a new fuzzy set which represents a combination of the old evidence and the new evidence based on how well the Matching fact matches the Premise. The AssertRule first calls GetCorrelationStrength to determine the strength of the match between the premise and the Matching facts. This method allows multiple premises to match with multiple matchings, e.g., if premise1 = matching1 and premise2 = matching2... The miminum correlation strength of all the premise-matching matches provides the strength for how the new evidence modifies the old evidence. Then, if the strength of the match exceeds the Threshold, it modifies the NewEvidence by the correlation strength and Infers a new fact (the result). |
Reference |