An Ensemble ML Classifier-based Fake Profile Identification Framework in Online Social Network

Abstract

Typically, online social networks include huge range of people all around the world and this becomes a huge part of their life. People exploit social networks for sharing their feelings so as to make friends, for setting up new businesses, thus to connect with family, friends & so on. Online social network offers huge benefits to individuals in varied ways but it too suffers with few limitations. There exist several people who exploit these networks for making harm to others on creating fake accounts on these networks. To detect such fake and genuine users, machine learning (ML) algorithm is used in this work. The ML models are employed for predicting and classifying datasets over three ensemble ML approaches employed. It becomes complex sometimes to differentiate among the outcomes of varied schemes and for this reason, an ensemble ML model is employed to make the task much easier and to enhance the accuracy. In this work, three different classification algorithms termed Random forest (RF), Logistic Regression (LR), and Support Vector Machine (SVM) are used by estimating their accuracy on Twitter dataset acquired from Kaggle repository. The results of these models attained are compared and is concluded that the best accuracy is attained from RF classifier model on comparing other two classifier employed.