Home > trafficking > Combating Human Trafficking with Big Data: Today’s Plot Twist

Combating Human Trafficking with Big Data: Today’s Plot Twist

When the government shuts down websites that serve the sex industry, the activity doesn’t stop—it just moves. All things considered, it’s constructive for the government to keep shutting things down. It does slow down illegal activity, at least for a while.  But on the other hand, it also slows down law enforcement, who rely on those sites for leads to help rescue victims of human trafficking.

Human trafficking—when the victim is forced to participate or is under age—is different from consensual sex work. Georgia Tech PhD student Julia Deeb-Swihart uses machine learning, network analysis, and techniques from information visualization to create tools to help law enforcement combat trafficking. As the activity moves, our research moves to follow it.

New legislation passed last month gives the government greater authority to prosecute websites that facilitate trafficking, and has the bad guys scared. But give them credit for being clever. Hiding in plain site is the Facebook group “best automotive reviews.” The main purpose of the page is to tell customers of a sex worker review site what the new URL is, each time the site moves. But in the mean time, people have been discussing their automotive needs, like these posts:

Screen Shot 2018-04-11 at 11.21.54 AMScreen Shot 2018-04-11 at 11.21.11 AMScreen Shot 2018-04-11 at 11.20.50 AM

It’s certainly a new challenge for machine learning to detect those posts! The technical challenge of the week. But wherever they go, we’ll keep chasing them.

Categories: trafficking Tags:
  1. Ali Heidari
    April 11, 2018 at 3:43 pm

    The problem is human creativity and mapping that creativity in ML algorithms. It is an interesting challenge. Interestingly, there is a limited set of figurative meaning/words (chick/car e.g., have another ferrari in garage, …) when reference is made to an object/human being and it could be there that one can decide what to feed the learning algorithm and find interesting patterns. If it was me I could have collected all possible references from AMT and use them to learn a model for prediction….

  1. No trackbacks yet.

Leave a Reply to Ali Heidari Cancel reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: