The goal of this research project is to enhance the identification and characterization of malware behaviors from Darknet data using near-real-time clustering with parallel computing and optimization, together with graph kernel-based clustering methods and machine learning.