FP-growth算法及其优化在税务系统中的应用  

Application of FP-growth algorithm and its optimization in tax system

作　　者：倪德 马传香 NI De;MA Chuanxiang(School of Computer Science & Information Engineering,Hubei University,Wuhan Hubei 430062,China;Education Information Engineering Research Center,Wuhan Hubei 430062,China)

机构地区：[1]湖北大学计算机与信息工程学院,武汉430062 [2]教育信息化工程研究中心,武汉430062

出　　处：《计算机应用》2018年第A02期140-143,共4页journal of Computer Applications

摘　　要：针对FP-growth算法中需要频繁建立条件模式(FP)树并递归挖掘的问题,提出了在单路径条件FP树情况下的一种基于组合思想而改进的FP-growth算法。通过组合单路径条件FP树中频繁一项集的方法,实现无需递归而达到挖掘频繁项集目的。并以某省地税局风险监控平台系统中的税务数据为基础,使用改进后的FP-growth算法,挖掘纳税人偷税漏税行为中频繁出现的税种税目之间的强关联规则,实验结果表明,改进后的FP-growth算法不仅提高了挖掘频繁项集的效率,而且为税务征管和稽查工作提供了更好的依据。In order to solve the problem that FP-growth algorithm needs to be established frequently and recursively,a new FP-growth algorithm based on combination was proposed.By combining frequent itemset on single-path conditional FP tree,frequent itemsets can be mined without recursion.Based on the tax data in the risk monitoring platform system of a local tax department,the improved FP-growth algorithm was used to find out the strong association rules between the frequent tax items appearing in tax evasion by taxpayers.The results show that the improved algorithm not only improves the efficiency of mining frequent itemsets,but also provides a better basis for tax collection and inspection.

关 键 词：数据挖掘 频繁项集 关联规则 FP-GROWTH算法 税务征管 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]