Fake Social Media Profile Detection Using Machine Learning and Deep Learning
P. Supraja, M Pranita, Gopika M Nair · 2025
The rapid evolution of social media networks has provided grounds for fake accounts, which pose risks like identity theft, online scams, and misinformation. The growing sophistication of fake accounts has made their detection with traditional techniques very difficult. To aid in this, we constructed a fully featured fake profile detection system that uses behavioural analysis based on machine learning, NLP based sentiment analysis on user bio and profile picture verification. The system employs a Random Forest classifier that was trained on behavioural features which included follower and following counts, the nature of the username, privacy of the account, and language misuse in the account's bio. It achieved an accuracy of 96%. For profile picture verification, the Mediapipe, Yandex Reverse Image Search, and DeepFace along with the ViT model provided the best results in recognizing altered and fake images achieving an accuracy of 96.5%. The combination of these techniques provides a complete and effective approach to determining the existence of fake accounts through behaviour analysis and image verification. The accuracy for detection of fake profiles through these means is 96.5%. This offers a complete method for detection of fake profiles through the synergy of the proposed techniques.