摘要
针对稠密数据集,提出一种基于单向FP-tree的最大频繁项集挖掘算法Unid_FP-Max2。该算法在挖掘过程中只生成被约束子树,而它是一种虚拟的树结构,在原有的单向FP-tree基础上用三个很小的数组来表示,因而避免了以往算法需递归构造条件FP-tree来计算最大频繁项集的弊端,极大的降低了内存空间和时间开销,提高了挖掘效率。实验表明,与FP-Max算法相比,算法的效率提高了1倍以上。
Proposes an efficient algorithm Unid_FP-Max2 for mining the complete set of maximal frequent itemsets in a unidirectional FP-tree. Because the algorithm only generates constrained sub-trees which is pseudo tree structure consisting of three small arrays based on the originally unidirectional FP-tree, avoides the flaw in former algorithms which need to generate lots of conditional FP-trees for finding maximal frequent itemsets recursively. Reduces the space and time consumption to a great extent,then the algorithm improves mining efficiency. Experiment shows that in comparison with FP-Max, this algorithm accelerates the mining speed by at least one times.
出处
《现代计算机》
2010年第1期19-24,共6页
Modern Computer
基金
河南省高校杰出科研人才创新工程项目(No2007KYCX018)