Honor Among Thieves: Analyzing Language Features of Darknet Market Vendors

Abstract

The World Wide Web enables retailers to provide goods and services to consumers around the globe, including illegal products and services. The digital black market (Darknet) includes digital exploits, hacker-for-hire services, drugs, weapons, and other illicit goods and services. Appropriately classifying threats on the Dark Web is critical to military and law-enforcement cyber surveillance, reconnaissance, and defense. We used statistical and sentiment analysis to create a unique digital profile of Darknet market vendors. We also identified characteristics that indicate truthfulness, deception, credibility, and intent by analyzing product description language. The features of these profiles were used with a recommender system to track vendors across marketplaces. Our experiment achieved a rank-1 vendor identification accuracy of 74.7%. We also used semantic fingerprinting to identify vendors that deviated from the market average. Many of these anomalous vendors were discovered to have indicators of fraudulent and deceptive practices. We concluded that using a recommender system and sentiment analysis on vendor language in Darknet marketplaces helps law enforcement and national security professionals to track and disrupt the sale of illicit goods and services.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 2020
Accession Number
AD1114715

Entities

People

  • John E Iii Smith
  • Nicholas E Hughes

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • Biomedical
  • Cyber

DTIC Thesaurus Topics

  • Artificial Intelligence
  • Blockchain
  • Commerce
  • Computational Science
  • Computer Languages
  • Computers
  • Cyberattacks
  • Dark Web
  • Data Analysis
  • Data Sets
  • Databases
  • Denial Of Service Attack
  • Electronic Commerce
  • Electronic Mail
  • Information Science
  • Internet
  • Language
  • Machine Learning
  • National Security
  • Natural Language Processing
  • Natural Languages
  • Network Protocols
  • Network Science
  • Spreadsheet Software
  • Supervised Machine Learning
  • United States

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Industrial Economics
  • Political Violence and Terrorism Studies.

Technology Areas

  • Cyber