M-distinct算法改进:面向动态数据集重发布的隐私保护算法  

Improvement of M-distinct:Towards Privacy Preserving Re-publication of Dynamic Datasets

作　　者：白似雪[1] 李婵[1] 

机构地区：[1]南昌大学计算机科学与技术系,江西南昌330031

出　　处：《南昌大学学报（工科版）》2010年第3期297-300,共4页Journal of Nanchang University(Engineering & Technology)

摘　　要：M-distinct extension算法通过将数据添加、删除、更新三种情况分开处理,在删除情况下引入伪记录的概念,使得攻击者无法得出插入和删除数据之间的联系;另一方面提出事实:数据记录的不同属性在不同的发布用途中其信息权重并不相同,根据事实,引入属性权重的概念,减少算法在分配和划分两个阶段的循环次数,提高算法的效率。M-distinct extension algorithm deals with the data add,delete,update three cases separately.With the introduction of the concept of false record in the delete case,the attacker cannot get the link between insert and delete data.On he other hand,the fact is presented：different attributes of data records in the release have different weights.According to the fact,the introduction of the concept of attribute weights reduces the number of cycles in the assignment and split of the two stages of the algorithm to improve the algorithm efficiency.

关 键 词：数据挖掘 隐私保护 动态数据集 M-distinct算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]