Efficiently Mining the Recent Frequent Patterns over Online Data Streams

In order to mine the frequent patterns in the recent time windows of the data stream than the historic, a new method was proposed to mine the recent frequent patterns in the sliding window of an online data stream. It calculated the approximate frequencies of patterns in the sliding window with a conservative strategy. Also, it built a Recent Frequent Pattern tree(RFP-tree for short) to incrementally capture the patterns in the sliding window and mine them by scanning the stream only once. Extensive experiments show that the proposed method is more efficient and scalable than other analogous algorithms.