Efficient Computation of Partial-Support for Mining Interesting Itemsets

Mining interesting itemsets is a popular topic in the data mining community. The objective of this problem is to mine all interesting itemsets, with respect to a given interestingness measure. While considerable efforts have being spent on justifying the various interestingness measures, the algorithms that mine them are not quite well-studied, except in the case support, which has resulted in the famous frequent itemset mining (FIM) problem. In this paper, we show that a certain class of interesting itemsets can be represented by functions of their partial support. This class includes some definitions of fault-tolerant itemsets, estimated support of itemsets in noisy data, and bond of itemsets. As the name implies, partial support of an itemset is the number of transactions containing some part of the given itemset. This paper addresses the problem of efficiently calculating partial supports, which leads to efficient algorithms for mining interesting itemsets in that class. We show that there exists a recurrence relation between partial supports. Hence, we can calculate the partial supports of itemset by simply extending any FIM algorithm (even the implementation). This allows us to benefit from innovations and optimizations in FIM algorithms. Theoretical analysis shows that our approaches retain the running time complexity of the base FIM algorithms for only a small cost in space. Extensive experiments on several real-world datasets also demonstrate that algorithms based on our approach are significantly faster than previously proposed techniques for corresponding definitions.