An Efficient Algorithm for Mining Erasable Itemsets

Mining erasable itemsets first introduced in 2009 is one of new emerging data mining tasks. In this paper, we present a new data representation called PID_list, which keeps track of the id_nums (identification number) of products that include an itemset. Based on PID_list, we propose a new algorithm called VME for mining erasable itemsets efficiently. The main advantage of VME algorithm is that the gain of an itemset can be computed efficiently via union operations on product id_nums. In addition, VME algorithm can also automatically prune irrelevant data. For evaluating VME algorithm, we have conducted experiments on six synthetic product databases. Our performance study shows that the VME algorithm is efficient and is on average over two orders of magnitude faster than the META algorithm, which is the first algorithm for dealing with the problem of erasable itemsets mining.