首页 | 本学科首页   官方微博 | 高级检索  
     

一种基于格的隐私保护聚类数据挖掘方法
引用本文:崔一辉,宋伟,王占兵,史成良,程芳权.一种基于格的隐私保护聚类数据挖掘方法[J].软件学报,2017,28(9):2293-2308.
作者姓名:崔一辉  宋伟  王占兵  史成良  程芳权
作者单位:武汉大学计算机学院, 湖北武汉 430072,武汉大学计算机学院, 湖北武汉 430072,武汉大学计算机学院, 湖北武汉 430072,武汉大学计算机学院, 湖北武汉 430072,武汉大学计算机学院, 湖北武汉 430072
基金项目:国家自然科学基金(61232002,61572378,61202034);CCF中文信息技术开放课题基金(CCF2014-01-02);武汉市创新团队项目(2014070504020237);武汉大学自主科研项目“面向保密数据的可信云数据共享服务系统”(2042016gf0020,2016-2017)
摘    要:由于云计算的诸多优势,用户倾向于将数据挖掘和数据分析等业务外包到专业的云服务提供商,然而随之而来的是用户的隐私不能得到保证.目前,众多学者关注云环境下敏感数据存储的隐私保护,而隐私保护数据分析的相关研究还比较少.但是如果仅仅为了保护数据隐私,而不对大数据进行挖掘分析,大数据也就失去了其潜在的巨大价值.本文提出了一种云计算环境下基于格的隐私保护数据发布方法,利用格加密构建隐私数据的安全同态运算方法,并且在此基础上实现了支持隐私保护的云端密文数据聚类分析数据挖掘服务.为保护用户数据隐私,用户将数据加密之后发布到云服务提供商,云服务提供商利用基于格的同态加密算法实现隐私保护的k-means、隐私保护层次聚类以及隐私保护DBSCAN数据挖掘服务,但云服务提供商并不能直接访问用户数据破坏用户隐私.与现有的隐私数据发布方法相比,论文的隐私数据发布基于格的最接近向量困难问题(CVP)和最短向量困难问题(SVP),具有很高的安全性.同时算法有效保持了密文数据间距离的精确性,与现有研究相比挖掘结果也具有更高的精确性和可用性.论文对方法的安全性进行了理论分析并设计实验对提出的隐私保护数据挖掘方法效率进行评估,实验结果表明本文提出的基于格的隐私保护数据挖掘算法与现有的方法相比具有更高的数据分析精确性和更高的计算效率.

关 键 词:数据挖掘  隐私保护  隐私保护的数据挖掘  基于格的加密
收稿时间:2016/7/10 0:00:00
修稿时间:2016/9/4 0:00:00

Privacy Preserving Cluster Mining Method Based on Lattice
CUI Yi-HUI,SONG Wei,WANG Zhan-Bing,SHI Cheng-Liang and CHENG Fang-Quan.Privacy Preserving Cluster Mining Method Based on Lattice[J].Journal of Software,2017,28(9):2293-2308.
Authors:CUI Yi-HUI  SONG Wei  WANG Zhan-Bing  SHI Cheng-Liang and CHENG Fang-Quan
Affiliation:School of Computer, Wuhan University, Wuhan 430072, China,School of Computer, Wuhan University, Wuhan 430072, China,School of Computer, Wuhan University, Wuhan 430072, China,School of Computer, Wuhan University, Wuhan 430072, China and School of Computer, Wuhan University, Wuhan 430072, China
Abstract:Due to the various advantages of cloud computing, users tend to outsource data mining task to professional cloud service providers. However, user''s privacy cannot be guaranteed. Currently, many scholars are concerned about how to protect sensitive data from unauthorized access, few scholars'' research on data analysis. But ifpotential knowledge cannotmined, the value of big data may not be fully utilized. This paper proposes a Privacy Preserving Data mining (PPDM) method based on lattice, which support ciphertext intermediate point and distance homomorphic computing.Meanwhile, webuild aprivacy preserving cloud ciphertext data clustering Data Mining Method. Users encrypt privacy data before outsource the data to cloud service providers, cloud service providers use homomorphic encryption to achieve privacy protection k-means, hierarchical clustering and DBSCAN mining algorithms. Compared with the existing PPDM method, our method with high security is based on shortest vector difficulties (SVP) and the closest vector problem (CVP). While it maintain the accuracy of distance between two data, therefore mining results with high accuracy and availability. We design experiments for the privacy preserving cluster mining (PPCM), real experimental data is cardiac arrhythmia datasets of machine learning, the experimental results show that the method based on lattice ensure not only security but also accuracy and performance.
Keywords:Data mining  Privacy Preserving  Privacy Preserving Data mining (PPDM)  Lattice-based cryptography
点击此处可从《软件学报》浏览原始摘要信息
点击此处可从《软件学报》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号