Research Of Association Rules Algorithm And Application In Data Mining

Nowadays,the capacity of people to grasp the data is rising.Facing so mass data,people more concern the very important information hidden in the data than the data itself.Data Mining meets our requirment,and it is a useful tool to help us find the important knowledge from the data.Association Rules are important branch of Data Mining,and excavating Association Rules of the large services Database plays an important part in the solving of actual problems of different domain.The paper mainly discusses Association Rules algorithm and its application.First,the paper sysmatically describes the related theory knowledge of Data Mining and Association Rules in order to lay the foundation for research content.Second,it indicates the deficiency of Apriori algorithm,that repeatedly scans Database and may generate huge candidate sets,and then we can find the entrance to new algorithm. So,the paper makes the following improvement to apriori algorithm:first,it gets L1, and L2 from the perspective of Database Boolean Matrix,breaking the natural mode of Lk generation in apriori;and then it improves the generation algorithm of Ck (k≥3)by proving conclusion,that Lk-1∞L1 can take the place of Lk-1∞Lk-1 to get Ck.So,we propose the improved algorithm,BMSL_Apriori algorithm (Boolean Matrix Simplified Linked_Apriori algorithm),by integrating above-mentioned work.First,we can know BMSL_Apriori algorithm not only reduces the number of Database scanning and avoid huge candidate sets generation,but also it drops the spending of time and space,by analysis of its theory. Second,we proves that the algorithm is superior to apriori and other algorithm by specific experiment. With some data of true transaction Database,we take advantage of Microsoft SQL Server 2005 and VB.NET to construct a simple Association Rules Mining system in good software and hardware environment,and we use BMSL_Apriori algorithm in the procedure of Association Rules Mining.Finally,it is verified that the it obtains the better effect than apriori and other algorithm by Mining result.