Efficient algorithms for mining recent weighted frequent itemsets in temporal transactional databases

Jerry Chun-Wei Lin, Wensheng Gan, Philippe Fournier-Viger, Tzung-Pei Hong

Abstract: In this paper, we present two efficient tree-based algorithms for mining recent WFIs (RWFIs), which consider both item weights and the recency of patterns. A novel tree called the recent weighted frequent (RWF)-tree and a sorted downward closure property of RWFIs are proposed. Moreover, two data structures, named element (E)-table and RWFtable, are designed to store the compact information. The RWFI-Mine and its improved RWFI-EMine algorithms are proposed to discover RWFIs in a recursive way and without candidate generation. Experiments show that the proposed algorithms outperform both the traditional WFIM algorithms and the state-of-the-art algorithm for mining RWFIs.