Research On Data Flow Association Rule Mining Algorithm Based On Sliding Window

As the database technology used in Government,business and other social organizations operating process widely,more and more kinds of data appears in the researchers ' view,such as spatial data,time series data,streaming data,multimedia data,text data,voice data,and so on.And streaming data due to the extensive business network transactions,stock,sensor networks,and many other fields,thus causing great enthusiasm of the researchers.Compared with traditional forms of data,the streaming data has a large,real-time,order,range of features and is not suitable for all of the data stored in the database,so the traditional data mining algorithm is not suitable for streaming data mining.And previous data mining algorithm is the default data items have the same importance and uniform distribution of each project,in practical application,however,is different and the importance of the items is evenly distributed.Based on this,we introduced the idea of multiple minimum support and weight.In practice,recent data on value,and interest.Therefore,the more common data stream mining methods are studied in a certain time period,this time we call the time window.In the study,the time window is divided into the landmarks window model/the attenuation window model/sliding window model,and the most widely used model is sliding windows.Association rules mining in order to find more than minimum support and confidence of all the rules that all strong association rules.Mining of association rules can be divided into two steps:the first step in finding the target on all frequent patterns in data;second,qualified with these frequent itemsets of association rules.Part ?dealing with simple and direct,decided by the first step in the overall performance of association rule mining.In real applications,if support is set too small,combinatorial explosion is thrown,causing a mining process can't proceed;if value is too large,you do not have access to rare items Association rules.Multiple minimum supports weighted approach is both the reality of the situation by the thought of a solution.Association rules mining in data streams,this article examines the sliding window of mining with multiple minimum supports based on matrix method of SWM-MMSW and top-k of frequent itemsets mining based on weight,to improve the algorithm,W-TKFM algorithm.These two algorithms using matrices and matrix to store transaction data and frequent 2-set.Verified by calculation and experiment,SWM-MMSW and W-TKFM algorithms to efficient mining of frequent item sets and have a good time and space efficiency.