Abstract :
There have been several algorithms for mining association rules, such as Apriori and some improved Aprioris, only to be Interested in those itemsets, which have support above a userdefined threshold. However, there exists a kind of important rule, indirect association, hidden in these itemsets, which are filtered out. When a pair of items, (A, B), which seldom. occur together in the same transaction, are highly dependent on the presence of another itemset, Z, the pair (A, B) are said to be indirectly associated via Z In this paper, the definition of indirect association is firstly given. Then a measure of dependence to estimate the correlation between relative frequent items and a simple way to express the closeness between a pair of items indirectly associated by another itemset are provided. In addition, two kinds of classifying standard for indirect association rules are proposed for further research. In order to mine such indirect association rules, an algorithm of indirect association mining (IAM) is presented. And the complexity analysis about this algorithm is showed. An experiment in order to verify the utility of this algorithm is made. Finally, some issues about the IAM algorithm are put forward for future research.
