@article{M337A097B,
title = "CAPTCHA Classification Framework for Dark Web Profiling",
journal = "The Journal of Korean Institute of Communications and Information Sciences",
year = "2025",
issn = "1226-4717",
doi = "10.7840/kics.2025.50.7.1118",
author = "Eunseon Yu, Gyuna Park, Seo-Yi Baik, Seongmin Kim",
keywords = "Dark Web, Crawling Bot, Bot Detection, CAPTCHA",
abstract = "Recently, criminal activities on the dark web, facilitated by anonymity through Tor, have exponentially increased. With the rise in illegal transactions, understanding the structure of black markets within the dark web ecosystem has become crucial for investigations. However, dark web operators actively minimize information leakage by restricting automated crawling bots. Among these measures, CAPTCHA is widely utilized and exists in various forms. This paper proposes an automated framework to identify and analyze the structure of black markets that use CAPTCHA to block or restrict access by crawling bots. It first introduces classification criteria for CAPTCHAs on the dark web and then presents a model capable of automating their classification. The CAPTCHA classification model collects over 120 links through crawling based on the jargon and terminology used in the dark web, then categorizes them according to the established CAPTCHA classification criteria. The proposed framework demonstrated a commendable CAPTCHA classification accuracy of 93.33% within the dark web. Based on this, it is expected to enhance the efficiency of dark web profiling through an automated crawling bot capable of bypassing CAPTCHA mechanisms."
}