Online data stream Mining of Recent Frequent Itemsets based on Sliding Window model

Abstract

Online data stream mining is one of the most important issues in data mining. Identifying the recent knowledge can provide valuable information for the analysis of the data stream. In this paper, we proposed an one-pass data stream mining algorithm to mine the recent frequent itemsets in data streams with a sliding window basing on transactions. To reduce the cost of time and memory needed to slide the windows, each items is denoted a bit-sequence representations. Basing on a priori property, this kind of representations can find frequent items in data streams efficiently. We named this method MRFI-SW (mining recent frequent itemsets by sliding window) algorithm. Experiment results show that the proposed algorithm not only attains highly accurate mining result, but also consumes less memory than existing algorithms for mining frequent itemsets over recent data streams.