Research on Multi-dimensional Association Rules Mining

The traditional Apriori algorithm is a common algorithm for finding Boolean association rules, but it is less efficient, and it is based on one-dimensional database, which can not be applied to multidimensional association rules data mining. In this paper, based on the traditional Apriori algorithm, we propose a new algorithm for multi-dimensional association rules data mining. The algorithm transposes and extends multi-dimensional sequence database, and uses a bitmap set to represent the transaction which used each item. It converts the sequence mining to the basic items mining, and then uses projection and bitwise operation to mine various dimensional frequent itemsets. Finally, by join operation it gets all the frequent itemsets. The new algorithm also resolves the problem of Apriori algorithm repeatedly scanning database and producing a large number of candidate frequent itemsets. Experiments prove that the algorithm can effectively complete the multi-dimensional sequence data mining.