fuzzy association analysis, t-norm, association rules

Fuzzy association analysis extracts relationships from data. The result of fuzzy association analysis depends on a chosen t-norm that is used for calculating confidence and support measures of mined association rules. We show that the set of mined association rules might change depending on the t-norm. We measure the distances of sets of mined rules with different t-norms and also with set of rules mined by crisp association analysis. We experiment with various datasets and partitioning methods to examine relationships of mined rules by different t-norms. Our experiments shed new light on application of fuzzy association mining and confirm that fuzzy association analysis usually brings signifficantly different results when compared to results given by crisp (non-fuzzy) association analysis.