基于敏感分级信息熵的匿名方法  

DATA ANONYMITY METHOD BASED ON SENSITIVE HIERARCHICAL INFORMATION ENTROPY

作　　者：石昆正 张攀峰 董明刚[1] Shi Kunzheng;Zhang Panfeng;Dong Minggang(School of Information Science and Engineering,Guilin University of Technology,Guilin 541006,Guangxi,China)

机构地区：[1]桂林理工大学信息科学与工程学院,广西桂林541006

出　　处：《计算机应用与软件》2024年第5期319-326,共8页Computer Applications and Software

基　　金：国家自然科学基金项目(61862019);广西自然科学基金项目(2017GXNSFAA198223);广西科技基地和人才专项(2018AD19136);桂林理工大学科研启动基金项目(GLUTQD2017065)。

摘　　要：针对相似攻击所造成隐私泄露的问题,提出(H,p,k)-匿名模型,通过对敏感属性分级,使等价类中元组不同敏感级别的个数满足设定阈值H,并设计满足该模型的匿名算法MAA-SLIE(Micro-aggregation Algorithm based on Sensitive Level Information Entropy)。该算法基于贪心聚类思想,在聚类过程中保证等价类隐私安全指数最大,提高等价类中敏感属性多样性,降低隐私泄露风险,减少信息损失,通过实验验证了算法的合理性和有效性。Aiming at the problem of privacy leakages caused by similar attacks,this paper proposes(H,p,k)anonymous model.By classifying sensitive attributes,the number of tuples with different sensitive level in equivalent classes could meet the set threshold H.An anonymous algorithm MAA-SLIE(microaggregation algorithm based on sensitive level information entropy)was designed to satisfy the model.Based on the greedy clustering idea,the algorithm ensured the maximum privacy security index of the equivalence class in the clustering process,improved the diversity of sensitive attributes in the equivalence class,and reduced the risk of privacy leakage and information loss.The rationality and effectiveness of the algorithm were verified through experiments.

关 键 词：数据匿名 信息熵 微聚集 隐私保护 

分 类 号：TP309.2[自动化与计算机技术—计算机系统结构]