Mapping the Underground: Supervised Discovery of Cybercrime Supply Chains Rasika Bhalerao∗, Maxwell Aliapoulios∗, Ilia Shumailov†, Sadia Afroz‡, Damon McCoy∗ ∗ New York University, † University of Cambridge, ‡ International Computer Science Institute
[email protected],
[email protected],
[email protected],
[email protected],
[email protected], Abstract—Understanding the sequences of processes needed expertise and connections. Machine learning has been used to perform a cybercrime is crucial for effective interventions. to automate some analysis of cybercrime forums, such as However, generating these supply chains currently requires time- identifying products that are bought and sold [26], however, consuming manual effort. We propose a method that leverages machine learning and graph-based analysis to efficiently extract using it to discover the trade relationship between products supply chains from cybercrime forums. Our supply chain de- has not been explored yet. tection algorithm can identify 33% and 42% relevant chains In this paper, we propose an approach to systematically within major English and Russian forums, respectively, showing identify relevant supply chains from cybercrime forums. Our improvements over the baselines of 11% and 5%, respectively. approach classifies the product category from a forum post, Our analysis of the supply chains demonstrates underlying connections between products and services that are potentially identifies the replies indicating that a user bought or sold the useful understanding and undermining the illicit activity of these product, then builds an interaction graph and uses a graph forums. For example, our extracted supply chains illuminate cash traversal algorithm to discover links between related product out and money laundering techniques and their importance to buying and subsequent selling posts.