基于差分隐私的数据匿名化隐私保护方法

Anonymized data privacy protection method based on differential privacy

在保护数据隐私的匿名技术中,为解决匿名安全性不足的问题,即匿名过程中因计算等价类质心遭受同质性和背景知识攻击造成的隐私泄漏,提出了一种基于差分隐私的数据匿名化隐私保护方法,构建了基于差分隐私的数据匿名化隐私保护模型;在利用微聚集MDAV算法划分相似等价类并在匿名属性过程中引入SuLQ框架设计得到ε-MDAV算法,同时选用Laplace实现机制合理控制隐私保护预算。通过对比不同隐私保护预算下可用性和安全性的变化,验证了该方法可以在保证数据高可用性的前提下有效地提升数据的安全性能。

There exists the problem of security insufficience among the data privacy protecting technology which is the privacy leakage caused by homogeneity and background knowledge attack when computing equivalence classes in the anonymity process. To solve the problem, an anonymized data privacy protection method based on differential privacy was put forward, and its model was constructed. ε-MDAV （Maximum Distance to Average Vector） algorithm was presented, in which micro-aggregation MDAV algorithm was used to partition similar equivalence classes, and SuLQ frame framework was introduced into the anonymous attribute process. Laplace mechanism was used to reasonably control the privacy protection budget. The comparison of availability and security under different privacy protect budgets verifies that the proposed method effectively improve data security while guaranteeing high data availability.