Behaviour Profiling of Reactions in Facebook Posts for Anomaly Detection

Malicious attackers are highly interested in Facebook which is the most widely used Online Social Network. Malicious activities are massive nowadays in this network and spreading fake news, sending spam messages, fake applications and like jacking are a few among them, which leads to huge financial and reputation loss. The current scenario in the world is that the malicious activities are executed by heavily funded criminal groups. A larger part of Facebook accounts are either fake or compromised and they take part in malicious activities. Finding these malicious accounts is a challenging task. Social influence based behavioural analysis is one of the approaches towards detecting malicious activities. Facebook users are influenced by other user's posts/reactions. On observing the change in reactions, anomalous behaviour of the corresponding accounts can be identified. This paper proposes a method based on unsupervised clustering which analyse the reactions of users called smileys. The reactions are profiled and by applying similarity measures and unsupervised clustering techniques, they are further classified. This approach reveals the behaviour of immediate emotional responses of users to the various posts in Facebook. Since reactions are immediate, the analysis of these reactions provides important information to find anomalous behaviour in Facebook accounts