基于邻居度序列相似度的k-度匿名隐私保护方案

摘要

Aiming at the problem of traditional k-degree methods which usually add a huge amount of random noise, have low success rate in re-constructing graph and cannot resist attacks based on more complex background knowledge of graph query, an advanced k-degree scheme is presented for preserving privacy. The scheme brings up the priority algo-rithm for constructing graph, adjusts the weights between neighborhood of nodes and the highest demand node by setting some parameters and introduces Euclidean distance to construct similar neighborhood degree sequence of nodes with the same degree after anonymity. Experimental results show that this scheme has less degree loss and can regulate collision between information loss and success rate of constructing graph, resist attacks about combining background knowledge of degree of nodes and their neighborhood degree sequence.%针对传统的k-度匿名方案添加随机噪声次数过多,构图成功率低,且无法抵御更复杂的图查询背景知识攻击的问题,提出了改进的k-度匿名隐私保护方案.该方案提出一种优先级构图算法,通过设置参数来调整邻居节点与度需求高的节点之间的权重,引入欧式距离并对k-度匿名后的同度节点构造出相似度较高的邻居度序列.实验结果表明,该方案的度信息损失较少,能够调节边信息损失与构图成功率之间的冲突,抵御以节点的度结合邻居度序列作为背景知识的攻击.