Research Of Pruning Redundant Association Rules Based On Sequential Patterns

Mining sequential association rules is an important research field of data mining, it reflects the interesting association or correlation between large amounts of itemsets. However, we usually get a lot of association rules, which contain a large number of redundant rules. It's difficult for users to analyze and use. A variety of techniques can effectively reduce the number of redundant rules. In view of on the deletion of redundant negative association rules algorithm is not a lot in our country, we proposed the following techniques for pruning negative redundancy pruning redundant rules.(1)This paper introduced the definition and the theorem of redundant rules, then we proposed the redundancy theorem in negative association rules. Specifically include the following: 1)association rules like (?)A(?)B : Given the rules of A(?)B and A(?) B', if A, B (?) I ,A∩B =Φ,B'(?) B,corrA,B<1, then we say the rule (?)A(?)B is the redundant rule of A(?) B'. 2) association rules like (?)A(?)B: Given the rules of (?)A(?)B'and (?)A(?)B, if A, B (?) I , A∩B =Φ,B'(?) B,corrA,B <1,then we say the rule (?)A(?)B'is the redundant rule of (?)A(?)B . 3) association rules like (?)A(?)B: Given the rules of (?)A(?)B and(?) A(?)B , if A, B (?) I,A∩B =Φ,B'(?) B,corrA,B >1,then we say the rule (?)A(?)B is the redundant rule of(?) A(?)B'.(2)When we minning negative association rules from sequential patterns ,we should also consider its subset, Because its subset can mine the rules which can't mine from itself. When we pruning redundant rules in the sequence, we must not only consider the time factor but also consider correlation. Because in the expression of correlation, supp (A∪B) express A occurred behind B. Combined with correlation and time factor, using the proposed pruning theorem of sequence positive and negative rules, the existing redundant association based on the pruning algorithm, an improved model based on sequential pruning of redundant rules, proposed a new technology of pruning redundant association rules based on sequential patterns and gave the algorithms. Experimental results show that the algorithm can effectively reduce the number of sequence rules...