基于动态规划的最大频繁项目集挖掘研究

将动态规划算法应用于最大频繁项目集的挖掘,可以克服Apriori算法需要多次扫描数据库确定新的候选项集的缺点;通过对数据进行初始化构建矩阵,结合动态规划的思想通过在矩阵中找到最大无向完全图来获得所有的最大伪频繁项集,最后利用一个非频繁项集的子集有可能是频繁项目集的性质对所有的最大伪频繁项集消减获取最大频繁项集。实验结果表明,它能够快速挖掘频繁项集,且适用于海量、高维数据。

Research of Mining Maximum Frequent Itemsets Based on Dynamic Programming

The application of dynamic programming to mining maximum frequent itemsets can overcome apriori algorithm＇s shortcomings of repeated scanning the database to determine the new set of options, through initialize data and build matrix. Combined with the idea of dynamic programming through found the largest free access complete graph in the matrix to get all of the greatest pseudo-frequent itemsets. Finally, we can make use of a non-frequent itemsets may have a subset of frequent itemsets to reduce all of the greatest pseudo-frequent itemsets to get the maximum frequent itemsets. The experimental results show that the algorithm can mining frequent itemsets rapidly, and can be applicable for massive, high dimensional data.