高效隐私保护频繁模式挖掘算法研究

阐述了隐私保护数据挖掘的目标,即在获取有效的数据挖掘结果的同时,满足用户对隐私保护的要求.针对个体用户及组织用户的隐私保护,论述了不同的方法,并归纳出隐私保护数据挖掘中所采用的两种主流算法.改进了高效隐私保护关联规则挖掘算法(EMASK)中需要完全的数据库扫描并且进行多次比较操作的弊端,提出了基于粒度计算的高效隐私保护频繁模式挖掘算法(BEMASK).该算法将关系数据表转换成面向机器的关系模型,数据处理被转换成粒度计算的方式,计算频繁项集变成了计算基本颗粒的交集.特别是数据的垂直Bitmap表示,在保证准确性不降低的情况下,一方面减少了I/O操作的次数,另一方面较大地提高了效率.

Research on Efficient Privacy Preserving Frequent Pattern Mining Algorithm

This paper elaborated the goal of privacy preserving data mining,that is to satisfy the demand of users for privacy protection as we acquire the mining results of effective data mining.For privacy protection to individual users and group users,this paper discussed different methods,and summed up two main algorithms in data mining of privacy preservation.Since efficient mining associations with secrecy konstraints(EMASK) needs full database scan and many comparison operations,the author came up with efficient mining associations with secrecy konstraints which is based on Bitmap computation(BEMASK).It transforms relational data form into relational model for machine,data processing is converted into granular computing method and calculation of frequent item-sets is turned into computing the intersection set of basic particles.Especially the vertical representation of Bitmap,under the condition of ensuring that accuracy is not reduced,on one hand reduces the number of I/O operations,on the other hand,greatly improves the efficiency.