Detecting Deceptive Fake Profiles in Online Social Media

Abstract

In Online Fake profiles present a security issue, online social networks (OSNs) help people communicate. Starting with data cleaning of the MIB dataset and manually collected data, this research use machine learning (ML) to identify fraudulent accounts. After testing a number of machine learning models, the Random Forest (RF) classifier showed the best accuracy. Max Voting (Majority Voting) increases cross-validation accuracy, whereas Extreme Gradient Boosting (XGBoost) and Decision Trees boost performance through ensemble learning. To ascertain validity, the algorithm examines profile details such as follower count, profile ID, and username. The RF classifier successfully identifies fake accounts, according to the results.