Collaborating with Local and Federal Law Enforcement for Disrupting Sex Trafficking Networks

• Published in: INFORMS journal on applied analytics 2022-09, Vol.52 (5), p.446-459
• Main Authors: Freeman, Nickolas K., Keskin, Burcu B., Bott, Gregory J.
• Format: Article
• Language: English
• Subjects: Advertisements; Algorithms; Circuits; Coercion; Data collection; Exploitation; Federal legislation; Filtering systems; Fraud; Human trafficking; Humans; Image classification; Internet; Law enforcement; Law enforcement agencies; Networks; Race; Sex; Teams; Trafficking; Victims; Websites
• ISSN: 2644-0865; 2644-0873
• DOI: 10.1287/inte.2022.1126

Sex trafficking, a form of human trafficking that involves sexual exploitation, has been facilitated through the use of online classified advertisements. Our research team has been collaborating with law enforcement agencies at the local, state, and national levels since May 2019 to detect and disrupt sex trafficking activities identified by applying advanced analytics to ad data collected from the internet. Human trafficking refers to the transportation, harboring, or obtaining of persons through force, fraud, and coercion for the purpose of exploitation. Every year, millions of individuals spanning all ages, races, genders, and nationalities are victims of human trafficking across the globe. Sex trafficking, a form of human trafficking that involves sexual exploitation, is facilitated through the use of online classified advertisements. Our research team has been collaborating with law enforcement agencies at the local, state, and national levels since May 2019 to detect and disrupt sex trafficking activities, which we identified by applying advanced analytics to ad data collected from the internet. Although off-the-shelf software exists to track and analyze individual cases, limited tools and analysis are available for linking information coming from multiple websites, constructing circuits with related ads, filtering networks based on geographical span, and predicting future movements. In response to these analytical gaps, our approach focuses on (1) developing models for predicting the future activity of selected networks for the purpose of interdiction planning; (2) creating unique mathematical and statistical algorithms for linking image and text data across sites into networks, each representing a set of connected text and image data; (3) establishing innovative metrics to classify and filter constructed networks on the basis of spatiotemporal advertisement patterns, text and emoji usage, and images; and (4) designing effective escort ads for use in operations targeting demand reduction. Our tool utilizes data collected from more than 10 ad sites and includes text and image details from more than 30 million ads. Using the tools we developed, the research team has assisted in several operations, which have led to more than 40 arrests in Tuscaloosa, Alabama, since February 2021. The research team continues to work with law enforcement partners in Tuscaloosa and Birmingham, Alabama, and Long Island, New York.