Dark web activity classification using deep learning

The present article highlights the pressing need for identifying and controlling illicit activities on the dark web. While only 4% of the information available on the internet is accessible through regular search engines, the deep web contains a plethora of information, including personal data and online accounts, that is not indexed by search engines. The dark web, which constitutes a subset of the deep web, is a notorious breeding ground for various illegal activities, such as drug trafficking, weapon sales, and money laundering. Against this backdrop, the authors propose a novel search engine that leverages deep learning to identify and extract relevant images related to illicit activities on the dark web. Specifically, the system can detect the titles of illegal activities on the dark web and retrieve pertinent images from websites with a .onion extension. The authors have collected a comprehensive dataset named darkoob and the proposed method achieves an accuracy of 94% on the test dataset. Overall, the proposed search engine represents a significant step forward in identifying and controlling illicit activities on the dark web. By contributing to internet and community security, this technology has the potential to mitigate a wide range of social, economic, and political challenges arising from illegal activities on the dark web.