Data mining technology has been maturating through over ten year's development, however, it is still a challenged work to mine data streams. The applications which show up these years have accelerated the proportion of data stream models. These applications include e-commerce, sensor network, and stock data analysis etc. Data in these applications is huge. It must be processed sequencely. However, traditional data mining algorithms can not process these huge, dynamic data.This paper researches frequent patterns mining in data streams. It begins with analyzing frequent patterns, then designs data structure and algorithms. Firstly, the paper designed a data structure CFIT in order to fit data streams which are huge and rapidly coming. This data structure stored summary of data streams. The structure of scan tree is reduced through adding super sets link into the frequent pattern trees.Then, this paper proposed an increment update algorithm TW-CFI based on frequent pattern trees and sliding windows. This algorithm adapted the characteristics of data streams.Secondly, this algorithm saved history information using inclining time windows, satisfying real time research researching afterwords. Non-frequent itemsets were deleted in order to improve mining efficiency through pattern pruning. It saved data storage and evaluated frequent closet itemsets better.Experimental results show that the algorithms proposed in this paper are more efficient than the current ones, and the anticipated results are realized. |