基于KFCMSA的(k,l)加权社交网络匿名算法

(k,l)weighted anonymous algorithm for social network based on KFCMSA

图数据隐私保护的研究目前主要集中在简单图,适应范围有限。将权重图数据的隐私保护作为研究对象,可以改善权重图发布之后数据的可用性及有效性。针对在利用聚类匿名化方法处理社交网络数据时,需要增删大量的边和节点,造成严重的数据失真的问题进行了研究。提出了(k,l)加权社交网络匿名算法KFCMSA(联合k成员模糊聚类和模拟退火),并利用改进的簇划分算法将权重社交网络聚类成不同的簇,对同一簇中节点的边权重进行泛化,使节点满足l多样性。在实现k度匿名的同时有效减少了边的改变量,提高了数据的可用性,实现最优聚类的同时防止了同质性攻击。聚类质量实验和数据可用性分析表明该算法具有较高的性能优势和较高的边保留率。

The general research of graph data privacy protection mainly focuses on simple graphs,which has a limited scope of application.Taking the privacy protection of the weight graph data as the research object can improve the availability and effectiveness of the data after the weight graph is published.This paper studied the problem of serious data distortion caused by the need to add and delete a large number of edges and nodes when using the clustering anonymization method to process social network data.It proposed the(k,l)weighted social network anonymity algorithm KFCMSA(combined k-member fuzzy clustering and simulated annealing),and clustered the weighted social network into different clusters using the improved clustering algorithm.It generalized the edge weights of nodes in the same cluster to make nodes satisfy l diversity.While implementing k-degree anonymity,it effectively reduced the amount of edge changes,improved the availability of data,and prevented homogeneity attacks while achieving optimal clustering.The clustering quality experiment and data availability analysis show that the algorithm has high performance advantages and high edge retention rate.