MDPA:基于MCL的社会网络差分隐私数据发布算法

摘要

The privacy protection of social network data publishing was to ensure the security of privacy information.In order to solve the problem of privacy protection in social network data publishing,a strict differential privacy preserving model was introduced.A new social network data publishing method based on Markov cluster algorithm was proposed,which was consistent with ε-differential privacy.It was named MDPA(MCL differential privacy algorithm).MDPA added noise which followed Laplace to network edge weights,and used Sias sampling frequency.The experimental results on real data sets showed that MDPA met user′s differential privacy requirements in social networks and improved the effectiveness of data.%社会网络数据发布的隐私保护是为了确保数据集中隐私信息的安全.针对社会网络数据发布所面临的隐私保护问题,引入严格的差分隐私保护模型,设计了一种基于马尔科夫算法(Markov cluster algorithm,MCL)并且满足ε-差分隐私的社会网络差分隐私数据发布方法.设计实现了满足ε-差分隐私(MCL differential privacy algorithm,MD-PA)算法,以Si为抽样频率,对网络边权重添加满足ε的隐私保护预算,服从拉普拉斯分布的噪声.真实数据集上的实验结果表明,MDPA算法满足用户在社会网络中的差分隐私要求,并提高了数据效用性.