Abstract
The spread of violent extremism and propaganda is a critical threat both nationally and globally. With the ever-increasing popularity and use of social media, spreading this extremism has never been easier for terrorist organizations and their followers. One terrorist organization, in particular, ISIS (the Islamic State of Iraq and Syria), uses Twitter for the vast majority of their social media interaction. It is crucial to have cyber analytics tools developed to combat these extremists' online presence and influence on social media platforms, such as Twitter. In this research, we apply machine learning algorithms to understand popular ISIS supporters' behavior and techniques and their possible influence on other users. We collected and analyzed a dataset containing over seventeen thousand tweets posted by pro-ISIS Twitterers. We utilized three machine learning algorithms with several models/settings in an attempt to classify and predict whether the top 4 pro-ISIS Twitter users (the most followed and tweeted users) authored a specific tweet. The algorithms applied in this work include sequential neural networks, random forests, and XGBoost. The models were ensembled, timed, and one model was simplified to attempt to improve performance and runtime.