Abstract: Existing anonymous technologies pay more attention to the utility of anonymous data, ignoring the problem that attackers recognize targets through a variety of background knowledge. In addition, with the scale of users increasing year by year, the traditional anonymous technologies can??t meet the actual demand. Motivated by this, a distributed k-degree-m-label anonymity method with protecting link relationships (called PLRD-(k, m)) is proposed. This method utilizes the message passing mechanism of GraphX, and divides nodes that are N-hop neighbors into one group and performs k-degree anonymous and m-label anonymous to ensure that attackers can??t identify targets by degrees and labels, and protects link relationships from being leaked. Finally, a protecting link relationships distributed personalized anonymity method extended from PLRD-(k, m) is proposed. Experimental results based on real social network datasets show that the proposed method can not only improve the execution efficiency of large-scale social networks, but also has good data utility.

Key words: social network, privacy preserving, distributed, k-degree-m-label anonymity, GraphX

摘要： 现有的匿名技术多关注匿名后数据的可用性，忽略了攻击者可以通过多种背景知识进行攻击的问题。此外，随着用户规模的逐年递增，传统的匿名技术已不能满足实际需求。为此，提出一种保护链接关系的分布式匿名方法PLRD-(k,m)(distributed k-degree-m-label anonymity with protecting link relationships)。该方法利用GraphX的消息传递机制，通过将互为N-hop邻居的节点分为一组并进行k-degree匿名和m-标签匿名，保证攻击者无法通过度和标签识别出目标并保护链接关系不被泄露。最后，扩展了PLRD-(k,m)方法，提出一种个性化匿名方法以满足用户不同的需求。基于真实社会网络数据集的实验结果表明，提出的方法不仅能提高处理大规模社会网络的执行效率，同时具有很好的数据可用性。

关键词: 社会网络, 隐私保护, 分布式, k-度-m-标签匿名, GraphX