摘要
隐私保密是数据挖掘应用中正迅速发展的研究领域。重点讨论了数据保密的两种方法:值—类成员关系、值变形,并在此基础上给出了保密程度量化的确定。数据挖掘的任务是研究并提出整合数据的模型,文中利用贝叶斯规则重构数据的原始分布,并将此方法应用于决策树的生成,得到了3种构建决策树的算法:全局算法、分类算法、局部算法,并且通过理论分析得出局部算法和分类算法比全局算法的效率高。
Privacy preserving is a new and rapidly emerging research area in data mining. Mainly discusses the two methods applied in data mining: value- class membership and value distortion, and forwards the quantifying privacy. The primary task in data mining is the development of models about aggregated data,so it makes use of Bayes' rules to reconstruct the original distribution. And it considers the three different algorithms that differ in when and how distributions are reconstructed: glohal, byclass, local. Finally, it analyses the execution time of three algorithms, and concludes that global's efficiency is the lowest,and byclass falls in between global and local.
出处
《微机发展》
2005年第10期4-5,25,共3页
Microcomputer Development
基金
湖北省教育厅重点科研项目(2003A002)
关键词
数据挖掘
随机化
重构
data mining
randomize
reconstruction