Research On Mining Closed Frequent Pattern In Data Streams

The advent of data stream triggered extensive research on data stream processing. Mining frequent patterns in data streams was a technology playing an important role in e-commerce and knowledge store applications. But as the dynamic and boundless characteristic of data stream, there was a redundant problem of data and items in all frequent patterns minging. It was high time and space complexity of the related mining algorithm. Pasquier proposed a concept of closed frequent pattern, which can determine exactly the complete set of all frequent patterns and was usually much smaller than the latter. Therefore it was a very important research on closed frequent pattern mining.At first, knowledge involved in data stream, such as the structure and model of data stream and the technology of data stream were introduced. Second, mining frequent and closed frequent pattern algorithm were studied, especially considering the problems such as the large searching space of the closed patterns and a bump when data update. Then, a new algorithm of mining closed frequent pattern in data stream with transaction sliding window was presented. A data structure BitVec-gen was designed for mining closed pattern directly without storing all itemsets through developed a concept of generator. Besides, by using bitwise AND to simplify accounting support number, through the left-shift operation to complete the items update in the sliding window.The algorithm compressed the transaction and items from horizontal and vertical direction to reduce memory by using binary bits representation of itemsets. Introducing a traditional algorithm to mine closed frequent pattern directly to reduce the searching space and improve the mining efficiency. Finally the example analysis and experiments were achieved, the experiments results show that the proposed algorithm not only attain accurate mining results, but also run significant faster and consume less memory than moment.