Abstract: Fuzzy association rules can be used to represent human knowledge in terms of natural language, and has attraded a growing amount of attention from the communities of Data Mining and Knowledge Discovery. However, so far, most approaches of mining fuzzy association rules arc based on the measures of support and confidence for classical association rules. From the viewpoint of fuzzy implications, fuzzy Horn clause rules, degree of support, implication strength and some related concepts were defined, and an algorithm was proposed for mining fuzzy Horn clause rules.This algorithm can be decomposed into three subprocess. First of all, a quantitative database is transformed into a fuzzy database. Secondly, all frequent itemsets in the fuzzy database that arc contained in a sufficient number of transactions above the minimum support threshold are identified. Once all frequent itemsets are obtained, the desired fuzzy Horn clause rules above the minimum implication strength threshold can be generated in a straightforward manner.

Key words: Fuzzy association rules, Fuzzy horn clause rules, Degree of support, Implication strength, Quantitative databases,Fuzzy databases