New algorithms for effectively creating conditional pattern bases of FP-Tree

The frequent pattern approach is undoubtedly a great milestone in the history of mining association rules,which greatly improved the mining effectiveness.However,Paths in Frequent Pattern Tree(FP-Tree) are repeatedly scaned in creating conditional pattern base,which results in great unnecessary performance overhead.In order to solve this problem,the information for creating conditional pattern base of the items related to each node of the path was computed and stored in the traverse of each path from leaves to the root,which avoided repeat scan of a single path.It is showed from the results of experiments that the revised ones obviously outperform the previous ones.