面向医疗数据可信共享的映射泛化(k,l)-匿名算法  

Mapping generalization(k,l)-anonymity algorithm for security sharing of medical data

作　　者：康海燕[1] 邓婕 KANG Haiyan;DENG Jie(School of Information Management,Beijing Information Science&Technology University,Beijing 100192,China;Computer School,Beijing Information Science&Technology University,Beijing 100101,China)

机构地区：[1]北京信息科技大学信息管理学院,北京100192 [2]北京信息科技大学计算机学院,北京100101

出　　处：《北京信息科技大学学报（自然科学版）》2021年第5期1-8,共8页Journal of Beijing Information Science and Technology University

基　　金：国家自然科学基金资助项目(61370139);教育部人文社科项目(20YJAZH046);北京信息科技大学科研水平提高项目(5211910933)。

摘　　要：针对医疗数据共享时(k,l)-匿名算法存在的敏感数据过度泛化问题和推理泄露问题,提出了一种基于映射关联原理和泛化边界值的泛化映射方法;建立了基于映射泛化的(k,l)-匿名模型,并形成面向医疗数据安全共享的泛化映射(k,l)-匿名算法。将本文算法和原始(k,l)-匿名算法、带映射关联的(k,l)-匿名算法以及带泛化约束的(k,l)-匿名算法在Kaggle医疗数据集进行对比实验,结果表明,本文算法不仅能保护医疗数据的安全,而且能够大大提高共享后数据的可用性。Aiming at the problems of over generalization of sensitive data and inference leakage in the(k,l)-anonymity algorithm in medical data sharing,a generalized mapping method based on the mapping association principle and generalized boundary value is proposed.A(k,l)-anonymity model based on mapping generalization is established and a generalized mapping(k,l)-anonymity algorithm for security sharing of medical data is formed.An experiment is conducted on Kaggle medical data set by comparing the proposed algorithm with the original(k,l)-anonymity algorithm,the(k,l)-anonymity algorithm with mapping association and the(k,l)-anonymity algorithm with generalization constraints.Experimental results show that the algorithm can not only protect the security of medical data,but also greatly improve the availability of shared data.

关 键 词：数据匿名化技术 (k l)-匿名 可信共享 映射泛化 

分 类 号：TP[自动化与计算机技术]