Mining interesting infrequent and frequent itemsets based on multiple level minimum supports and minimum correlation strength

摘要

Infrequent itemsets become very important because there are many valued negative association rules in them and multiple level minimum supports (MLMS) model can be used to mine infrequent and frequent itemsets. Some of the itemsets discovered by MLMS model, however, are not of interest. So in this paper, we propose a new model IMLMS to prune those uninteresting itemsets by improving Wu's pruning method (Wu et al., 2004), a method for pruning uninteresting itemsets. The shortcoming of IMLMS model is that the interesting measure minimum interest is greatly influenced by the support of corresponding itemsets, which adds difficulty for users to give a suitable value. So, we propose a new measure, minimum correlation strength (MCS), as a substitute to minimum interest to improve the performance of IMLMS model. The experimental results show that the IMLMS model works well and the MCS method has better performance than minimum interest.