Abstract

Underground forums contain many thousands of active users, but the vast majority will be involved, at most, in minor levels of deviance. The number who become engaged in serious criminal activity is small. That being said, underground forums have played a significant role in several recent high-profile cybercrime activities. We have compiled a massive dataset, dubbed CrimeBB, by crawling and scraping an assortment of online forums. The dataset presents a unique opportunity to understand these communities at scale, and allows for longitudinal social data analysis. Manual analysis is infeasible, and the complexity of these forums, and the unique lexicon used, makes automatic analysis challenging. In this talk we will describe the data collection and present preliminary results obtained in the scope of an interdisciplinary project, where we apply various data science methods to analyse the data. Concretely we apply social network analysis to analyse their social interests, natural language processing to classify the type of information posted and clustering to group the actors based on forum activity.