Predictive Patterns of Sex Trafficking Online
In the past 10-15 years, the internet has become a popular tool for sex traffickers to advertise and sell their victims. For instance, there are thousands of posts each month selling sex on the common classifieds website Backpage.com, some of which may be cases of sex trafficking. I extracted this publicly available data from the U.S. cities represented on Backpage.com. I then used software developed at Carnegie Mellon’s Auton Lab to find out if it is possible to detect patterns emerging from the data available in sex ads, such as patterns of travel that traffickers may use. To help identify which posts are more likely to be cases of trafficking, I relied on guidance from law enforcement experts. The research in this paper shows that it is in theory possible to track the movement of similar posts—and therefore, similar pimps or victims—across the country over time. With further development and refinement, the techniques demonstrated in this thesis could become the foundation for a valuable tool for law enforcement to use to prosecute traffickers and rescue victims.
Advisor(s)Jay Aronson, Artur Dubrawski