A decremental approach for mining frequent itemsets from uncertain data

Abstract

We study the problem of mining frequent itemsets from uncertain data under a probabilistic model. We consider transactions whose items are associated with existential probabilities. A decremental pruning (DP) technique, which exploits the statistical properties of items' existential probabilities, is proposed. Experimental results show that DP can achieve significant computational cost savings compared with existing approaches, such as U-Apriori and LGS-Trimming. Also, unlike LGS-Trimming, DP does not require a user-specified trimming threshold and its performance is relatively insensitive to the population of low-probability items in the dataset. © 2008 Springer-Verlag Berlin Heidelberg.