Detecting Slang on the Dark Web Based on Word Co-Occurrence Relationships in Anchor Texts
Abstract The dark web is widely used for illegal activities such as drug trafficking and cybercrime, where specialized slang is often employed to conceal criminal intent. We propose a novel method for detecting crime-related slang by leveraging the hyperlink structure of dark web pages. By analyzing HTML hyperlinks and identifying words that share common link destinations, our method discovers terms closely associated with criminal activities. Experiments on a large-scale dataset demonstrate that the proposed approach effectively detects a wide variety of crime-related slang while reducing noise.
Authors Ken Shiozawa, Hiroo Hayashi, Soramichi Akiyama, Marie Katsurai
Publication venue ICOIN 2026
Reference
Ken Shiozawa, Hiroo Hayashi, Soramichi Akiyama, and Marie Katsurai. 2026. Detecting Slang on the Dark Web Based on Word Co-Occurrence Relationships in Anchor Texts. In 2026 40th International Conference on Information Networking (ICOIN), pp. 347–352. doi: 10.1109/ICOIN68469.2026.11480587.