Reinforcing and Detecting Clone Profiles in Online Social Networks

Abstract

Online Social Networking (OSN) has brought real lives of people into the virtual world, lending itself to a wide variety of uses, but at the same time leaving its users vulnerable. OSN users make a great deal of personal information public in their profiles, often neglecting security precautions. An attacker takes advantage of the fact and creates clone profiles. When these clone profiles become active, the attacker sends friend requests to the victims’ friends, inviting them to join the clone’s network. If they do so, the attacker is in a position to cause annoyance and mischief to the genuine user. It is extremely difficult for other users to identify clone identities, as most of the attribute values are identical between genuine and clone profiles. This paper focuses on and seeks to contribute to the solution of the problem of profile clone attacks. It does so in two steps. First, it presents a potential form of clone attack called ‘reinforced snowball sampling clone attack’. The experimental results of reinforced snowball sampling performed on Facebook showed that in-out friend networks, with the highest acceptance level for friend requests, are more susceptible to attack when request message is sent in native language. The paper goes on to present a novel framework of a clone recognizer called Clone Detector which depicts three effectual clone identification techniques. The Clone Detector uses a combination of content related and content free techniques to efficiently detect clone accounts on various social networks.

Keywords: Content free techniques, Content related, Online social network, Profile clone attack, Reinforced snowball sampling clone attack.

View PDF