基于用户信息的社交网络信任评估方法

User information based trust evaluation mechanism for social network

社交网络的信任评估是社交网络应用安全的核心组件,公平客观的信任评价结果对于用户获取信息的正确性非常关键。利用用户之间信息的相似度进行用户聚类,实现用户群体的划分,依据用户集群的结果对用户之间关系进行修正,调整信任评估结果;同时还考虑到了恶意用户的影响,加入了信任检测的方案来保证方案的鲁棒性。经过实际社交网络数据实验仿真证明,该算法不仅可以使得信任评估结果与同类用户预期更为一致,而且可以大幅度降低恶意用户刷分行为的影响,实现有效而可靠的信任评估。

Trust evaluation is a key part of social network security. An impartial and objective way to evaluate trust is impor- tant for users to acquire requisite information. This paper utilized data clustering method to classify user based on users＇ back- ground information, and corrected relationship between users by their background information to estimate trust value. In addi- tion, it introduced behavior tracking mechanism to withstand malicious acts. After simulation test, this proposed method can not only get trust value closer to users＇ expectation, but also restrict the malicious effects of malicious users, thus achieving ef- fective trust evaluation.