Fast algorithm to find Minimal Generator

Abstract

Number of frequent closed itemsets (FCI) is usually fewer than frequent itemsets. However, it is necessary to find Minimal Generator (mG) for mining association rule from them [3],[5]. Finding mG approach based on the method of generating candidate is very time-consuming when number of frequent itemsets is large. In this paper, we present MG-CHARM, an efficient algorithm to find all mG of frequent closed itemsets. Based on the considering of mG features mentioned in 2.4 section we develop an algorithm which do not generate candidate by mining directly mG of closed itemsets at the same time of generating. Thus, the time for finding mG of closed itemsets is insignificant. Experiment shows that the time of MG-CHARM mining method is significant fewer than the time of finding mG after finding all closed itemsets (CHARM), especially in case the frequent itemsets number is large.