Abstract:
Social networking sites have turned out to be extremely well known as of late. Most internet users utilize them to discover new companions, refreshes their current companions with their most recent feelings and thoughts. Among these sites, Twitter, the quickest developing networking site, additionally pulls in numerous fake users to penetrate genuine users’ accounts with a lot of fake news, malware, viruses, and so on. This paper identifies effective features to distinguish fake accounts from legitimate accounts. Firstly, 20 features from users’ tweet content and user’s profile, which check possibly, are extracted to distinguish fake accounts. Effective features are selected from these 20 features using two feature selection methods. Validations of the effective features are proofed on Decision Tree method. The experimental results of five machine learning classifiers are shown in this paper and Random Forest classifier achieves the best detection accuracy with 95.7%.
