Deeply Understanding Structure-based Social Network De-anonymization

An Approach to Prevent Neighborhood Attack over Social Media

“…[11,12] The neighborhood attack based on a user and its neighbor's information identifies the isomorphic structure. [13][14][15] If two or more neighborhood networks are isomorphic in the social network graph, then the adversary cannot place a unique vertex neighborhood sub-network. [16,17] The proposed methodology increases the isomorphic neighborhood network in the social network graph by adding established imitation relationship edges.…”

“…The neighborhood attack is predicated on the idea that if an attacker gathers information about the victims' node's neighbors and their relationships, they can re-identify the victim's node from an anonymized social network. [15,16] Suppose if an attacker knows that A has five friends, three of V's friends {u1,u2,u3} are friends with each other, {u4,u5} are connected and the last node is only the friend of V, Figure 2 represents the 1-neighborhood network of vertex V. An attacker can use this graph to identify a since 1-neighborhood graph is unique to each social network node. The information of vertex V's friends and the relationship between them is the background knowledge of the attacker, based on background knowledge attacker make a 1-neighborhood network of vertex V and use this graph to identify vertex V in published social network.…”