The Studying Of Sequence Pattern Mining Based On DF2Ls

Sequence pattern mining is refer to mine the higher frequency sequence pattern than others in relative time.It has been widely used in production and people's life.It is the further promotion and application for association rule mining,the important direction of the data-mining. Sequence pattern mining algorithm can be divided into the method created based on the candidate code of Apriori and the other based on pattern grouth. However, the traditional methods all have some technical flaws and shortages,such as Apriori mode algorithm and sequence pattern gouth mode,they can make a huge candidate set and a big cost of mining system.Based on this, we study two kinds of algorithms based on the introduction of two new pruning strategies: DSEP (dynamic order extended cut) and DIEP (dynamic project extension pruning). With a view to enhance the efficiency of Apriori-like algorithm by pruning methods when the frequent sequential pattern mines.This paper proposed two pruning methods,DSEP and DIEP which can be used in all Apriori-like sequence mining algorithms or lattice-theoretic approaches, we optimize SPAM by using proposed pruning strategies and present the improved algorithm, SPAM+.The core is to use of DSEP and DIEP to delete the search space of spam,to share the 2-sequences dynamic frequency list (DF2Ls). Finally,through many experimengts,the comprehensive analysis result show that the performance of the SPAM+better than SPAM.