Mining Recent High Expected Weighted Itemsets from Uncertain Databases

Abstract

Weighted Frequent Itemset Mining (WFIM) has been proposed as an alternative to frequent itemset mining that considers not only the frequency of items but also their relative importance. However, some limitations of WFIM make it unrealistic in many real-world applications. In this paper, we present a new type of knowledge called Recent High Expected Weighted Itemset (RHEWI) to consider the recency, weight and uncertainty of desired patterns, thus more up-to-date and relevant results can be provided to the users. A projection-based algorithm named RHEWI-P is presented to mine RHEWIs based on a novel upper-bound downward closure (UBDC) property. An improved algorithm named RHEWI-PS is further proposed to introduce a sorted upper-bound downward closure (SUBDC) property for pruning unpromising candidates. An experimental evaluation against the state-of-the-art HEWI-Uapriori algorithm is carried on both real-world and synthetic datasets, and the results show that the proposed algorithms are highly efficient and acceptable to mine the required information.