摘要
某些情况下提取关联规则挖掘时需要根据项目的特点设置不同的最小支持度,针对此问题进行了多重最小支持度的频繁项集挖掘算法研究。在FP-growth的基础上提出了多重最小支持度树(MS-tree)的新方法,并设计了MS-growth算法对MS-tree进行频繁模式集的挖掘。该算法只需扫描一次数据库,克服了MSapriori算法在生成关联规则时需要重新扫描数据库的缺点。实验表明,新算法的性能可以和FP-growth算法相比,而且可以处理多重最小支持度的问题。
Mining frequent itemsets algorithm based on multiple minimum supports was studied in this paper, because sometimes setting different minimum supports to mine frequent itemsets is necessary. A new Minimum Support tree （MS-tree） algorithm and a MS-growth algorithm to mine all frequent itemsets based on Frequent Pattern growth （FP-growth） were proposed. It solves the problem of MSapriori algorithm that it cannot generate association rules without scanning the database again. The experimental results show that the proposed algorithm is comparable to FP-growth algorithm, but the former can solve the problem of multiple minimum supports.
出处
《计算机应用》
CSCD
北大核心
2007年第9期2290-2293,共4页
journal of Computer Applications
基金
国家科技支撑计划项目(2006BAF01A46)
上海市社会发展重大专项项目(06DZ12001)
上海市基础研究重点项目(06JC14066)
上海市科技发展基金重点项目(061612058)
上海市登山行动计划项目(061111006)
