Language and Sentiment Analysis of Extremist Behavior in Online Game Communities

Language and Sentiment Analysis of Extremist Behavior in Online Game Communities was a Multimedia, Hypertext, and Information Access capstone project to assist the VT Gamer Lab in gathering more data to analyze links between online video game communities and extremist behavior. Specifically military simulation games (referred to as milsims) were analyzed due to the inherently political and violent nature of the gameplay. The deliverables for the project were a community forum and YouTube scraper, cleaned data, visualizations, and sentiment analysis. We collected large datasets from both the community forums and YouTube, successfully cleaned the data, and did an analysis to create interesting visualizations. Sentiment analysis was originally going to be conducted with the client but was delayed past the submission of our report so we created our own analysis methods to produce interesting visualizations. Two of these visualizations showed us that the potentially extremist language on both platforms is very similar, both in word choice and frequency. This suggests that the communities define the language more than the platform they exist on.