k-匿名机制下查询隐私的一种度量方法

A privacy preserving measurement for query under k-anonymity mechanism

在k-匿名机制下,提出一种以信息熵和对数函数为基础的查询隐私度量方法.首先,建立k-匿名机制下的查询隐私的度量模型,包含4种角色和4种操作,为隐私保护的度量提供形式化的描述.然后,介绍两种背景知识的量化方式.针对第二种方式,由于用户属性离散化后的数值会被计算入背景知识概率表达中,造成背景知识概率表达的不准确,为此提出将离散化后的用户属性值作为特定查询和用户属性相关量的下标来求得相关量,从而进一步得到用户发出此特定查询的概率,摆脱了用户属性离散化后的数值对量化结果的影响.最后,提出查询隐私的度量方法.实验结果表明,该隐私度量方法能够较为有效地度量k-匿名机制下查询隐私算法的保护水平.

A query privacy measurement was proposed under the k-anonymity mechanism.The method was based on information entropy and logarithmic function.First,a framework for query privacy under the k-anonymity mechanism is established,which contains four roles and four operations provides a formal description for privacy measurement.Then,two quantitative methods of background knowledge are introduced.For the second step,user attribute discretization values will be calculated as a probability expression of background knowledge,affects the accuracy of the probability expression.The value of each user attribute after discretization was proposed as the index of the array to calculate the relevant quantities,the index of the array being generated by the relevance of the particular query and the attributes of the user,so as to further obtain the probability of the user issuing the particular query,thus avoiding the influence of discretized values of user attributes on the quantification results.Finally,a query privacy measurement is proposed.The experimental results show that the method can effectively measure the level of protection of the query privacy protection algorithm under k-anonymity mechanism.