Research On Frequent Pattern Mining Algorithm Of Data Stream Based On Sliding Window

With the explosive development of Internet technology,a large number of users have accumulated a large amount of data on the platform,and the amount of real-time data has grown rapidly in the form of an index.Most of generated data active in various fields in the form of data streams.Data stream is different from static database.In addition to the large amount of data,it also has the characteristics of infinite fastness,uncertainty,time-varying,and short-lived evanescence.Traditional data mining techniques can not met existing requirements,so we need to improve algorithms and related processing techniques that are more suitable for data flow characteristics.In order to obtain the availability and relevance of these data knowledge,more attention has been paid to frequent pattern mining of data streams.In dynamic data stream mining,for massively high-speed data stream mining,the algorithm should minimize the space complexity,Increase its adaptability to face the dynamic changes of data,improve the real-time processing data,The data flow should be quickly entered to avoid channel congestion,ensure that the data stream is processed in real time,and extract effective knowledge from it.Currently,the result of data stream mining is an approximation within the error range.In order to adapt to the characteristics of the data stream to mine the data stream better,this paper firstly deals with the data stream and data stream processing system.In-depth analysis is carried out on the existing theoretical basis,mining algorithm and utilization technology of frequent data pattern mining and frequent closed pattern mining.This paper analyzes the three result set models generated by mining,selects the frequent closed mode which takes up less space and can reflect the global by local.Based on the analysis of classification and clustering mining algorithms,this paper mainly improves the FP-growth algorithm and proposes the VSFP algorithm.The research questions mainly focus on the following two points:First,for the improvement of FP-tree,the prefix tree is used to propose a pattern tree that only needs to be traversed once,and the data is better compressed,and the data can be quickly performed;Secondly,comparing the three window technologies,the final selection is more suitable for the sliding window of the data stream,and it is improved,so that it can adaptively change the window size according to the sliding condition of the data stream,thereby saving system memory space.The VSFP algorithm uses three kinds of simulation data sets to conduct experiments,and compares the time and space efficiency of the results under different conditions.And analyze the results.The experimental results show that the VSFP algorithm has good time and space efficiency in the frequent mining of data streams.