Abstract
Frequent itemset mining is a typical question in the data mining domain. There are many researches about this problem already. We can use frequent closed itemset to find all frequent itemset, and the number of frequent closed itemset is much smaller than frequent itemset. Because of these strong points, frequent closed itemset mining becomes an important topic recently. In former researches, we can separate the approaches by data format into two ways: horizontal format and vertical format. And they can be transformed to each other easily. However, no matter what kind of format we use, we have to check if the answer is "closed". And that is overhead. Our new approach uses both horizontal format and vertical format data at the same time. One characteristics of this approach is that we do not need to check “closed” again. Moreover, in certain special dataset our approach has extremely good performance.