Application of AI Techniques to Deep Web Social Analysis

Application of ai techniques to deep web social network analysis

Online Social Networks (Twitter, Facebook, Telegram, and so on) have taken a huge place in the informational space, and are often used for propaganda, manipulation, or even recruitment by authoritarian states or terrorist organisations. As the quantity of information makes difficult the human exploitation, solutions to support the military decision makers can only come from the use of AI techniques to extract intelligence from posted texts, to qualify the users’ behaviours, and to identify the emerging social structure. In this article, we will illustrate how to use such techniques on a very peculiar social network hidden in the Dark Web. Thanks to the use of TOR, its members have an excessive feeling of freedom, enabling unconventional posts including hate speech and other hacker offers. We propose to unfold an analysis of 1000 days of activities of this network using NLP techniques to find the most interesting emerging topics. On selected areas of interest, we explain how these techniques can also support the discovery of key actors of the network. We then proceed with a ML-based profiling of the user behaviors. Finally, we introduce influence and cohesion scores for groups of users, which help their characterization and evaluation.

Download PDF