密集型数据最大频繁模式挖掘方法研究  

Research on Method of Mining Maximum Frequent Patterns of Dense Data

作　　者：何昀[1] 张继夫 闫彬 HE Yun;ZHANG Ji-fu;YAN Bin(Aviation University of Air Force,Changchun Jilin 130021,China)

机构地区：[1]空军航空大学,吉林长春130021

出　　处：《计算机仿真》2022年第10期435-439,共5页Computer Simulation

基　　金：2021年军队双重项目子课题(SZ059-BZ11-JG11-10)。

摘　　要：采用目前方法挖掘最大频繁模式时,没有对密集型数据进行预处理,无法消除密集型数据中存在的噪声,导致方法存在去噪性能差、挖掘效率低和挖掘准确率低的问题。提出密集型数据最大频繁模式挖掘方法,采用曲波变换方法对密集型数据进行稀疏描述,依据压缩感知理论,通过OMP算法消除密集型数据中存在的噪声,实现数据的去噪处理。在此基础上,建立分布式窗口树,通过更新分布式增量、分布式剪枝处理和频繁模式输出三个部分完成密集型数据最大频繁模式的挖掘。仿真结果表明,所提方法的去噪性能好、挖掘效率高、挖掘准确率高。When mining the maximum frequent patterns, the current method often ignores the preprocessing for dense data and is unable to eliminate the noise in dense data, leading to poor denoising performance, low mining efficiency, and low mining accuracy. In this article, a method for maximum frequent pattern mining in dense data was proposed. The curvelet transform was introduced to sparsely describe the dense data. According to the compressed sensing theory, the noise in dense data was eliminated by the OMP algorithm, so the noise was removed. On this basis, a distributed window tree was constructed. Finally, the maximum frequent pattern of dense data was mined by updating the distributed increment, pruning, and outputting frequent patterns. Simulation results show that the proposed method has good denoising performance, high mining efficiency, and high mining accuracy.

关 键 词：密集型数据 最大频繁模式 压缩感知理论 数据挖掘 分布式窗口树 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]