A Generalized Proposed Approach to Identify the social media Fake Accounts using SVM-NN: Machine Learning Perspective

Abstract

The social network platforms Google+, Facebook, Twitter are surprisingly consideration in the most recent decade pertained to the combination of social behavior and computational systems. Leading organizations, industries, and promoters use fake social media accounts to promote services and social activities like marketing, political movements, and sale promotions. Recently, many fake accounts created on social media platforms. This is the main securities challenge for the users that attracted cybercriminals and attackers to carry malicious activities, steal personal data and information, share false news, and request money. Several techniques and algorithms were proposed to detect fake accounts and check users' behavioral, non-behavioral activities and characteristics on social media networks.  The main contribution is machine learning, support vector machine (SVM), K- nearest neighbor (k-NN), decision tree, Neural Network, Artificial Intelligence, Convolutional neural network, Naive Bayes to detect fake accounts.  This paper proposed a generalized approach to detect fake accounts using a SVM and NN. Our proposed system uses fewer features and steps to identify the account's identity (real or fake) compared to other methods with higher accuracy. To evaluate the results using our proposed approach, we used the Twitter dataset available on (S. Cresci et al., 2015) and found our approach outperformed.