A new validation method of fuzzy association rules based on the Structural Equation Modeling

The fuzzy association rules have been introduced in order to get useful and interesting information from numeric databases. They offer to the user an understandable representation of knowledge. However, the task of mining fuzzy association rules generates a huge number of rules; most of them are redundant and uninteresting. To give a solution to this problem, we suggest a validation method that uses statistical tools, such as Structural Equation Modeling (MES). Our method consists in: (i) extracting rules based on the closure of the Galois connection that provides a generic basis of all association rules. (ii) evaluating and ranking these rules through the coefficient of MES in the objective of presenting them to the expert for validation.