基于差分隐私的高效用项目集挖掘算法  

An Efficient High Utility Itemset Mining Algorithm Based on Differential Privacy

作　　者：马煜 荀亚玲[1] MA Yu;XUN Ya-ling(School of Computer Science and Technology,Taiyuan University of Science and Technology,Taiyuan 030024,China)

机构地区：[1]太原科技大学计算机科学与技术学院,太原030024

出　　处：《太原科技大学学报》2023年第6期491-497,503,共8页Journal of Taiyuan University of Science and Technology

基　　金：国家自然科学青年科学基金(61602335);山西省自然科学基金(201901D211302);太原科技大学博士科研启动基金(20172017)。

摘　　要：在大数据时代,互联网安全事件频繁发生,用户数据安全性已成为大数据分析的热门话题。通过有效挖掘高效用项集获得的知识可能包含重要信息,如果被恶意滥用,可能威胁到数据拥有者的隐私或利益。为了防止数据信息泄露,提出了基于隐私保护的高效用项目集挖掘算法DPUP-Growth(based on Differential Privacy Utility Pattern Tree).首先在构建树的过程中,使用指数机制来混淆项头表的顺序,将拉普拉斯噪声添加到每个节点,从而得到差分隐私的树结构DPUP-Tree.最终进行高效用项目集挖掘。该方法以牺牲部分完整性为代价,大大提高了用户数据的安全性。实验结果表明,该方法的完整性损失在误差可接受的范围内,安全性能大大提升。In the era of big data,Internet security incidents occur frequently,and user data security has become a hot topic of big data analysis.Knowledge gained through effective mining of efficient item sets may contain important information.If maliciously misused,it may threaten the privacy or interests of the data owner.In order to prevent data leakage,an efficient item set mining algorithm DPUP-growth(based on Differen tial Privacy Utility Pattern Tree)is proposed.First,in the process of Tree construction,the index mechanism is used to confuse the order of item head table,and The Laplacian noise is added to each node,so as to obtain the DPUP-Tree with differential privacy.Finally,efficient project set mining is carried out.This method greatly improves the security of user data at the cost of sacrificing partial integrity.Experimental results show that the integrity loss of the method is within the acceptable range of error,and the safety performance is greatly improved.

关 键 词：高效用项目集挖掘 隐私保护 差分隐私 拉普拉斯机制 指数机制 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]