Classification of malicious and legitimate nodes for analysing the users´ behaviour in heterogeneous online social networks

With exponential growth of Internet in the modern era, social networking has evolved as a preference for the users. The communication overload has exposed insecurity in the network. To cope with this problem, the unwanted users´ behavior needs to be distinguished from legitimate users´ behavior to make social networking safe for users. In this paper, an analysis has been performed on a subset of two real datasets, one of Facebook links and other of Live Journal by applying classification algorithms. This paper employs REP Tree, Naïve bayes Multinomial Updateable, Complement Naïve bayes and Classification via Clustering algorithms for the classification of the datasets in online social networks like Hi5, LinkedIn, Facebook, Twitter, Live Journal etc. Moreover, on the basis of parameters like TP rate, FP rate etc., the best algorithm for the classification of malicious and legitimate users for analyzing users´ behavior in heterogeneous online social networks is selected.