云计算下权重社会网络差分隐私保护数据聚类方法

Weighted social network differential privacy protection data clustering method under cloud computing

传统方法在对权重社会网络数据进行保护时,未进行差分隐私数据的节点密度计算,数据保护效果不佳。为此,提出云计算下权重社会网络差分隐私保护数据聚类方法。分析差分隐私数据特征,确定正交多项式回归系数的特征序列;根据特征序列实现数据降维;采用加权共协矩阵实现差分隐私保护数据融合;根据互信息和分形维数完成网络差分隐私保护数据聚类过程。实验仿真证明,在IK数据集下此方法的聚类正确率为87.32%,迭代次数为28次;在UC数据集下聚类正确率为83.37%,迭代次数为32次,迭代次数减少,正确率明显高于传统方法。所提方法具有较高的正确率而且耗时较少,为实现社会网络下的各种差分隐私数据保护提供了有效的理论依据。

When the traditional method protects the weighted social network data,the node density calculation of the differential privacy data is not performed,and the data protection effect is not good.To this end,the introduction of cloud computing technology,clustering protection of differential social data of weighted social networks.To this end,a weighted social network differential privacy protection data clustering method is proposed under cloud computing.Analyzing the characteristics of differential privacy data and the sequence of features of orthogonal polynomial regression coefficients are determined.Data dimensionality reduction is realized based on feature sequences,and weighted co-join matrix is used to realize differential privacy protection data fusion.The network differential privacy protection data clustering process is completed according to mutual information and fractal dimension.The experimental simulation proves that under the IK data set,the clustering accuracy rate of this method is 87.32%,and the number of iterations is 28 times.Under the UC data set,the clustering correct rate is 83.37%,the number of iterations is 32,and the number of iterations is reduced.The rate is significantly higher than the traditional method.The proposed method has higher accuracy and less time,which provides an effective theoretical basis for realizing various differential privacy data protection under social networks.