用二分图实现数据发布的隐私保护*

基于表存储而发布的数据虽然可以实现隐私保护，但是由于表中记录相互独立，使得个体间的关联信息在发布中缺失，影响发布数据的效用。提出采用二分图的形式对数据进行发布，将顶点划分为两类，把带有标签的顶点按聚类方法进行分组，根据聚类分组结果对另外一个顶点集进行最大匹配分组，通过隐藏个体和顶点的映射关系，保证两类个体间关系的安全发布。基于聚类的最大匹配分组方法既实现了隐私的保护又增加了发布数据的效用。

Privacy-preserving data publishing using bipartite graph

It could implement privacy protection based on the table storage and data publication, but the records were independent each other. It made entities relationships miss in the publication and influenced the effectiveness of the publication data. With bipartite graph publishing data, divided the vertexes into two categories. Grouped the vertexes with a label by clustering method. Another vertex set implemented maximum matching group according to it. By hiding mappings between individual and vertex, it ensured relationships between two classes of individual security release. The maximum match group based on the cluster not only realizes the privacy protection but also increases the published data effectiveness.