Spammer Detection and Fake User Identification on Social Networks

Abstract

This study explores various methods for detecting Twitter bots and spam content, specifically focusing on analyzing Twitter data to identify fake accounts, spam URLs, and trending topics. The research utilizes multiple machine learning algorithms, including Naive Bayes and Random Forest, to classify tweets and user behavior. The analysis involves importing JSON-formatted tweets from multiple users and processing the dataset through a series of steps, including detecting fake content and assessing account authenticity. A key approach in the study is the application of Extreme Machine Learning (EML) algorithms, which improve upon traditional methods, offering an accuracy of 87.5% in identifying fake accounts compared to 50% accuracy achieved by Random Forest. Furthermore, a prediction accuracy