OSINT-Based Tool for Social Media User Impersonation Detection Through Machine Learning
Rajaa Alqudah, Mohammed Al-Qaisi, Rakan Ammari, Yazan Abu Ta'a · 2023
With the explosion of social media platform usage in recent years, privacy and online security have become major concerns. Malicious users create fake social media profiles, posing as regular people or public figures to gather personal information, damage reputations, or show off their social engineering skills. As a result, social media platforms, such as Facebook, Instagram, and others, provide the perfect place for identity theft. Malicious users can create fake social media profiles with the same name and profile image as another person. In this paper, Open Source Intelligence (OSINT) platform is used to collect and analyze publicly available information to identify potential impersonators. The proposed approach employs web scraping, machine learning, and web development modules in Python and can be hosted on the AWS cloud for optimal performance and scalability. It accurately scrapes social media platforms (Facebook and Instagram) and presents potential impostor profiles based on a user's uploaded photo only, or the photo with a name, or the photo with the name and a username. The proposed model achieved an accuracy of 88% with precision and recall matrices of 86% and 89% respectively.