Fake Social Media Profile Detection

Summary

Social media like Twitter, Facebook, Instagram, and LinkedIn are an integral part of our lives. People all over the world are actively engaged in these social media platforms. But at the same time, it faces the problem of fake profiles. Fake profiles are generally human‐generated or bot‐generated or cyborgs, created for spreading rumors, phishing, data breaching, and identity theft. Therefore, in this article, we discuss fake profile detection models. These differentiate between fake profiles and genuine profiles on Twitter based on visible features like followers count, friends count, status count, and more. We form the models using various machine learning methods. We use the MIB dataset of Twitter profiles, TFP, and E13 for genuine and INT, TWT, and FSF for fake accounts. Here, we have tested different ML approaches, such as Neural Networks, Random Forest, XG Boost, and LSTM. We select significant features for determining the authenticity of a social media pro file. As a result, we get the output as 0 for real profiles and 1 for fake profiles. The accuracy achieved is 99.46% by XG Boost and 98% by Neural Network. The fake detected profiles can be blocked/deleted to avoiding future cyber‐security threats.