大数据中的分类隐私加密方法研究与仿真

摘要

This article proposes a method for privacy encryption of classification in large data based on modified.Firstly,the author used data set in current time window to adjust data weight in other window based on Bayesian theory and structured Bayesian classifier according to different window weight,then built a classification model of Bayesian multi-window data flow to acquire privacy data waiting for encryption.Moreover,the theory of optimum bucket partition was used to extract value of partition and switch SQL statement querying database.Finally,the privacy encryption was completed according to the extracted results.Simulation results prove that the method can improve whole performance of system of classification privacy encryption in big data.%在大数据中蕴含了很多丰富的信息,是具有较高价值的数据集合,为了避免由于数据挖掘而造成的用户隐私泄露,需要进行大数据中的分类隐私加密,但是采用当前算法进行数据分类加密时,难以提取需要加密的数据特征,存在大数据中的分类隐私加密误差大的问题.提出改进最佳桶划分的大数据中的分类隐私加密方法.在叶贝斯理论基础上,采用目前时间窗口上的数据集调整其它窗口上的数据权重,根据不一样窗口权重构造贝叶斯分类器,构建贝叶斯的多窗口数据流分类模型,获取需要加密的隐私数据.利用最佳桶划分理论提取加密数据的划分值和转换查询数据库的SQL语句,利用提取的结果完成对大数据中的分类隐私加密.仿真结果表明,所提方法能提高大数据中的分类隐私加密系统的整体性能.