Human trafficking is big business. Now worth more than the international arms trade, the recruitment, transport and harbouring of people is a $150 billion per year ‘industry’ that trades on exploitation and misery of over 40 million people. Tracking the movement and exploitation of the victims of this crime is challenging but new techniques that use data and analytics are helping to break up slavery rings and call the perpetrators of these crimes to account.
At this year’s Splunk.Conf event in Las Vegas, Sherrie Caltagirone, the Executive Director of the Global Emancipation Network, talked about how data from a variety of sources was being used to track down exploited people and break down the ‘pipeline’ of slaves. In one case, a network that forced women out China, through New York and eventually to Florida to work in illegal brothels was broken.
The search for the illegal operations starts by ingesting unstructured text from a variety of sources. These include product reviews from Yelp and Rubmap – an online review service for legitimate and illegal massage parlours, information from listings of registered therapists and arrest and conviction data from public records. Using that data, the Global Emancipation Network was able to build a list of establishments that they believed were most likely to be harbouring trafficked people that were being exploited.
The data was ingested into a data lake and then overlaid using Google Maps to narrow down the potential locations.
A lexicon of keywords was built by the network in order to track specific terms that pointed to a high likelihood of illegal sex trade. For example, in mosts contexts, the word ‘fresh’ would have specific connotations. But in the sex trade, it is a key word that very meaningful. Even the way emojis are used in reviews is analysed.
As well as automating the process of text analysis, there was a very direct human benefit. Caltagirone said that reading the comments and reviews from sites known to harbour exploited people was traumatic and content moderators found the comments disturbing. In one case that Caltagirone highlighted, the reviewer realised that he was paying for sex from an exploited person and apologised for his actions. But he didn’t abstain from the activity or report the illegal brothel.
That analysis allowed the network to build a list of establishments that were most likely to be harbouring slaves. The model assigned high, medium and low risk to each establishment that was analysed. In Florida, where the Global Emancipation Network focuses, there are about 22,000 known facilities offering massage services. About 2,500 of those had online reviews with almost 20 per cent of those identified as being likely to be either operating as illegal brothels or engaged in human trafficking.
Direct investigations then followed in order to establish whether the automated analysis was accurate. As well as identifying human trafficking, arrests were made for the crimes including money laundering.
The success experienced by the Global Emancipation Network is driving an expansion of its activities. The network plans to look at other jurisdictions said Caltagirone as well looking at other verticals outside the massage industry. Aviation and social media are also areas that will come under the network’s scrutiny as it seeks to break up more people trafficking rings. And there will be moves to look at data in other languages as the network is currently focussed on English.
Caltagirone said the Global Emancipation Network is not working alone. Stop the Traffik is also disrupting the illegal slave trade by using analytics to track financial transactions and identify patterns by working with banks, information from survivor narratives and news reports to build predictive models through the application of artificial intelligence and natural language processing.
The Global Emancipation Network also works with Seattle Against Slavery. It uses chatbots to intercept conversations from people who are trying to procure sex from minors and other exploited parties. They use AI to detect when someone is trying to access a service from an exploited person. The conversation is intercepted and the potential procurer of services is redirected to education about child exploitation and offered support services to break them out of a damaging behavioural cycle and reduce the risk of recidivism.