Deep Learning Approach for Identification of Fake Profiles in Social Media

Abstract

Fake profiles have been developed as social media usage has increased, which can result in identity theft, cyberbullying, and online fraud. In order to secure internet users, efficient phony profile detection systems are required. In this study, we assess the efficacy of three well-known machine-learning algorithms and deep learning methods for the detection of fake profiles such as Support Vector Machines (SVM), Random Forest, and Neural Networks.The dataset aided for training and testing includes several variables taken from social media profiles, such as status_ count, followers count, friends count, favorites count, and listed count. The testing results demonstrate that all three algorithms are capable of identifying phony profiles with a high degree of accuracy, with neural networks having the best accuracy (99.2%). This study implies that machine learning algorithms have the potential to identify fraudulent profiles.