Research On Mining Maximal Frequent Patterns Over Data Stream

The association rule mining is a very important problem in data mining. The issue of mining frequent patterns plays a crucial role in association rule mining,sequential pattern mining, etc. Mining frequent patterns over data streams is a key issue on study of mining frequent patterns. Because of the time-consuming in mining frequent patterns, mining maximal frequent patterns has been proposed to improve the mining efficiency. The set of maximal frequent patterns contains all sets of the frequent patterns. The set of maximal frequent patterns is orders of magnitude smaller than the set of frequent patterns and there are applications where the set of maximal frequent patterns is adequate.In some applications, because the speed of data streams is non-constant, how to mine frequent patterns in this kind of data streams is an issue worth studying; Mining results is better of the new transactions and litter of the old transactions is also an interesting issue. In all, it is very significative to do some researchs on those issues. In this paper, we have done some researches on the related problems of data stream mining. It is stated as follows:(1)A new algorithm, BFPM-Stream, for mining maximal frequent patterns over data streams was proposed. It used transaction and time sensitive sliding window to resolve the unfavorable effects caused by the unsteady speeds of data stream. In addition, the methods, which were the bit objects for data expression and bit frequent patterns tree, BFP-Tree, to store and handle the data, were used. The experimental result verifies the efficiency of the BFPM-Stream.(2)A new algorithm based on damped transactions in data streams, BFPMW-Stream, for mining maximal frequent patterns was proposed. The transaction sliding window was adopted in the algorithm. In addition, the methods, which were the bit objects for data expression,bit frequent patterns tree, BFP-Tree, and storing patterns tree, P-Tree, to store and handle the data, were used. It could mine maximal frequent patterns in data streams that were made up of new or old transactions. The experimental result verifies the efficiency of the BFPMW-Stream.