Abstract: Linking spam on social networks is a process of constructing links among unrelated users to gain the influence. With the extensive application of social networks, linking spam would seriously threaten the information security of social networks. To tackle this problem, this paper proposes four types of users’ features and estimates relationship strength with a graphical model. Further more, propagating algorithm based on both trust and distrust with relationship strength is proposed to combat linking spam on social networks. Experimental results show that the proposed scheme can demote spammers and penalize users that link to spammers effectively.

Key words: Key words: social networks, linking spam, trust propagation； distrust propagation