数据发布中维护敏感数据高可用性的隐私保护方法.pdfVIP

计算机研究与发展 ISS N 100O一1239l CN l l 一1 l T P Journ aI of Cb l11Puter Res ae rch and Development 44 (Suppl . ) : 2 14 一2 19 ，200 数据发布 中维护敏感数据高可用性 的隐私保护方法 王雅哲 杨晓春 王 斌 于 戈 ( 东北大学信息科学与工程学院 沈阳 1100 04 ) ( yZa h司122@ 163 .Cor 。) A Technique of r Pr eser ving H igh Usability of Sen sit ive Dat a in Dat a P ublishing Wang y za he ，Yang Xioa chun ，Wa眼 Bin ，and y u Ge ( 六田l of l刀of rmatio n ience and Eng ￡nee r ￡ng ，NO r t为ae srte n Un i二 ￡yt ，hS e儿，以ng l l 0004) Abstract Trade一off between safety and usability is the main concern in privacy preserving data publishing pro blems 。Recently there presented a no vel technique一anatomy of r publishing sensitive data ，which can assure the safety and accuracy of published data 。But the disadvantage of anatomy is that it rn ay great ly distubr t he asos ciat ion and distribut ion inof rm at ion in the microdata 。To os lve this p roblem ，a new method called ClassAnatomy is presented ，w hich can preserv e the safety and accuracy in j ust the same way as anatomy . And meanwhile ，ClassAnat omy can preserve more inof rmation in the microdata by dividing the dat a space acco rding to t he class inof rmat ion of the quasi一ident ifier attribut es . Two different ClassAnat omy algorithms are given : the toP一down single dimensional division algorithm一T DA and the bottom一uP mult i- dimensional division al即rithm一BUA . Experiment re sults show that without co mpro mising saef ty ， Clas As natomy can dramatically reduce t he inof rmation loss in published data and rn ake the data more u sefu l . Key word s data publishing ; dat a privacy ; dat a usability ; anat omy method ; 1一diver sity 摘 要 数据 的安全性和可用性是敏 感数据发布与共享环境 中面临的主要 问题. 近期提 出了一种利 用 有损连接保证敏感数据发布的安全性的方法一Ana to my ，其优点在于发布的数据 中保留了原始数据的