基于k-匿名的多源数据融合算法研究  

Research on Data Fusion Algorithm for Multi-party Based on k-anonymity

作　　者：杨月平[1] 王箭[1] 

机构地区：[1]南京航空航天大学计算机科学与技术学院,江苏南京210016

出　　处：《计算机技术与发展》2017年第5期102-107,共6页Computer Technology and Development

基　　金：中国博士后科学基金(2014M561644);江苏省博士后科学基金(1402034C)

摘　　要：数据在当今的网络环境下变得越来越重要,融合技术能够使不同数据提供者有效地融合他们的数据,并且提供给顾客可定制且有效的服务。数据融合技术通常采用每轮自顶向下选择候选者,并进行数据更新的方法,而这种方法随着数据量的增加使得数据融合的时间花费巨大,难以满足数据融合的时间需求。为了减少融合数据过程中的花费,提高多源数据融合的精度,结合自顶向下分类树算法TDS,属性分类树,提出了一种基于k-匿名的多源数据融合算法。利用GUI的Adult数据集进行仿真实验,并比较了数据融合的复杂度以及融合精度的差异。实验结果表明,所提出的基于k-匿名多源数据融合算法融合过程时间花费更少,可以达到理想的数据融合精度,同时还实现了多源数据的融合。In today＇ s network environment, data has become more and more important. Data integration technology can make the effective data integration for different data providers, and provide customized service for the customers. Data fusion technology usually adopts the top-down to choose candidates for updating data in each round, and with the increase of amount of data, this kind of method costs a lot of time, which is difficult to meet the time requirements of data fusion. In order to reduce the cost in the process of data fusion and improve the accuracy of data integration for multi-party, a multi-party data fusion algorithm based on k-anonymous combining with the top -to-down TDS algorithm and the attribute classification tree has been proposed. Simulation experiments have been conducted with Adult set of GUI as well as comparison of accuracy of data fusion with complexity. The experimental results show that the proposed algorithm has taken less time and effectively achieve ideal accuracy of data fusion.

关 键 词：数据融合 K-匿名 自顶向下分类树 属性分类树 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]