Research On Publishing Algorithm Of Differential Privacy Protection For Social Networks

With the advent of the era of big data,a huge amount of information will be generated on social networks every day.By mining and analyzing these data,user experience and service quality can be well improved.However,these data often contain a large amount of personal privacy information,such as personal information,social relations and so on.Therefore,in order to prevent the disclosure of users' privacy information,it is necessary to protect the privacy of these data before it can be released to the third party for research.Differential privacy has been widely studied and applied in recent years because it has a strict mathematical theoretical basis and does not depend on any background knowledge of attackers.There are two kinds of differential privacy protection models in social network:node differential privacy protection and edge differential privacy protection,and the former can provide higher privacy protection intensity than the latter.Therefore,this paper makes use of the node differential privacy protection model to realize the privacy protection of social network data.This paper proposes a SNE(Sequence of Nodes and Edges)social network graph processing algorithm because the application of node differential privacy protection tends to bring high global sensitivity.This algorithm mainly constructs a sort rule combining nodes and edges to realize the stable sort of nodes and edges in the graph and improve the stability of the algorithm.At the same time,according to the threshold of node degree,the ordered edges are inserted into the "initialized" social network graph in turn,and the processing of the social network graph is finally completed.By setting a threshold value for the degree of nodes,the SNE algorithm makes the processed graph satisfy node differential privacy protection with a lower controllable global sensitivity,which greatly improves the availability of data and the performance of the algorithm.Aiming at the processed social network graph,this paper proposes a histogram data publishing algorithm based on k-means.The algorithm firstly performs initial transformation to the histogram,which avoids the problem that clustering the histogram interval directly leads to the destruction of the differential privacy protection mechanism.Secondly,based on the idea of partition plus noise,the histogram is partitioned by k-means algorithm and Laplace noise is added to each interval.Finally,the noise of each partition is evenly distributed to the histogram interval contained in the partition,and the order of the interval is adjusted to complete the publication.The threshold mechanism in the SNE algorithm avoids the influence of outliers on partition results in histogram partitioning,optimizes the partitioning effect,and then reduces the added noise in each interval of the histogram,and avoids the problem of data availability reduction caused by excessive accumulation of noise in large range count query.In this paper,the proposed algorithm is tested on real data sets.The experimental results show that the proposed algorithm can reduce the global sensitivity and noise accumulation error of the algorithm and improve the availability of the data on the premise of ensuring the privacy of social network data.