%0 Journal Article
%T 基于动态区间树的差分隐私数据发布算法<br>Differential privacy data publishing algorithm based on dynamic interval tree
%A 李丽
%A 张琳
%A 王汝传
%J 南京邮电大学学报（自然科学版）
%D 2017
%X 数据发布中的差分隐私是近几年隐私保护的热门研究，目前的数据发布研究中，主要有两种发布思路，第一种是先转换后添加噪声，另一种是先添加噪声再进行转换。文中在数据发布方面提出了一种新的思路，在原来的第一种思路基础上，先对数据源进行转换，然后添加噪声并转换为区间树，最后再进行数据的发布。在对直方图进行转换的过程中，为了提高查询精度，提出一种新的基于任意结构的区间树构造方法（CRTree算法），该方法将直方图转换为伪完全k叉区间树，并通过实验对比分析了该算法的误差来验证算法的效率，实验表明了算法的可行性和有效性。<br>The application of differential privacy in data released is topical issues of privacy protection in recent years. In the resent studies of the data publishing, there are mainly two lines of thinking. The first one is to add noise after the conversion and the other is to add noise firstly and then convertion. A new way in the field of data publishing is proposed based on the first idea for convering the primary data firstly, then add noises and converse to an interval tree to finish data publishing. In the process of histogram transformation, a new method is presented to construct the interval tree based on arbitrary structure (CRTree algorithm) to improve the query precision. The method transforms the histogram into similar complete k interval tree and verify the efficiency by comparion analysis of errors of the algorithm. Experimental results show that the algorithm is feasible and effective
%K 差分隐私 直方图 区间树 数据发布&lt
%K br&gt
%K differential privacy histogram interval tree data publishing
%U http://nyzr.njupt.edu.cn/ch/reader/view_abstract.aspx?file_no=201704017&flag=1