移动社交网络中隐私保护的用户相似度计算协议研究

Privacy Protected User Proximity Computing Protocol in Mobile Social Networks

在移动社交网络中朋友推荐的过程中,为了提供更好的服务,通常需要计算两个用户之间的相似度值。为了避免在相似度计算过程中泄露用户的属性等隐私信息,设计了基于Paillier加密算法的三方点积计算协议。利用两个用户隐私属性向量的点积值来代表用户间的相似度,对两个用户间的隐私向量进行加密后再计算其点积值,然后判断点积值是否满足系统或用户设定的某个阈值的要求。在此过程中,保证用户的隐私信息和点积值不被泄露给任何一方。性能分析和实验测试结果表明协议具有较高的安全性和运行效率。

During the process of friend recommendations,in order to provide better service,it usually needs to compute the similarity value of two users.In order to avoid the leakage of user＇s privacy information during the process of calculating user＇s similarity,a three party dot product calculating protocol which is based on Paillier encryption algorithm is set up.The dot product value of the private attribute vectors is used to manifest the similarity of the two users.The privacy vectors of two users will be encrypted before calculating the dot product value,and then,estimating whether the dot product value meets the threshold requirement which is set by the system or the user.In the process,the proposed protocol can ensure the user＇s privacy information and the dot product values will not be leaked to any party.Performance analysis and experimental results show that the protocol has high security and efficiency.