Enhancing Social Media Integrity: A Multi-Platform Machine Learning Approach to Detecting Fake Accounts with a User-Friendly Streamlit Interface
R. E., S Anunandhana., M Subhashree. · 2024
One of the greatest impacts of social media existence is the arising of fake accounts, which is, in many cases, related to extreme and negative actions such as the propagation of fake news or even fraud. This paper proposes the machine learning method in identifying fake users on social networking sites including Twitter, Facebook and Instagram. Utilizing such algorithms as Random Forest, AdaBoost, and XGBoost the system reaches, respectively, 97% of accuracy of fake accounts identification of Facebook, 100% of fake accounts identification of Twitter, and 91% of false accounts identification of Instagram. For the purpose of interaction with the user and making predictions in real-time, there is a simple Streamlit application. To sum up the key contributions, the proposed work underlines the importance of the ensemble strategies in general and Random Forest in particular when it comes to fake accounts detection, which gives an opportunity to provide efficient solutions for social media accounts’ enhancement.