Reproduce “Automatic User Profiling in Darknet Markets: a Scalability Study” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Claudia Peersman, Matthew Edwards, Emma Williams, Awais Rashid
Researchers at the University of Bristol developed an automated approach to user profiling in Darknet Markets using natural language processing and machine learning. Their method, analyzing individual messages, infers demographic characteristics like age and gender from textual communications, achieving improved performance compared to traditional user-level analysis.
Key insights
- Message-level analysis consistently outperformed traditional user-level analysis, indicating that individual posts reveal subtle linguistic cues for demographic prediction and deception detection.
- Character n-grams proved most effective for both age (59.5% micro F-score) and gender (63.8%) classification, while chatspeak features were particularly useful for the 25-34 age group (67.7% accuracy).
- Naive Bayes classifiers, specifically Complement Naive Bayes, achieved the best overall performance, and cross-domain transferability was limited for age prediction but more robust for gender classification.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.