Clustering Frequent Itemsets Based on Generators

How to reduce the number of frequent itemsets effectively is a hot topic in data mining research. Clustering frequent itemsets is one solution to the problem. Since generators are lossless concise representations of all frequent itemsets, clustering generators is equivalent to clustering all frequent itemsets. This paper proposes a new algorithm for clustering frequent itemsets based on generators. Firstly, based on minimum description length principle, the rationality of clustering generators is discussed. Secondly, the pruning strategies and mining algorithm for generators are proposed. Finally, based on a new similarity criterion of frequent itemsets, the clustering algorithm is presented. Experimental results show that the proposed method can not only reduce the number of discovered itemsets, but also is efficient.