Human trafficking is a widespread and devastating criminal enterprise that spans continents and crosses the world. Exact numbers of those affected and the organizations involved are difficult to come by, but the footprints of this activity can be seen online — and analyzing the patterns in that activity gives law enforcement agencies the tools they need to stop these criminal networks.

To do this, the founders of Marinus Analytics created Traffic Jam, a tool designed to assist law enforcement in identifying and taking down human trafficking organizations by sifting through digital information gathered from the online activity of these groups.

Over the past year, Marinus worked with Amazon.com Inc. and used the cloud-based graph database Amazon Neptune to boost the efficiency of Traffic Jam and help automate tedious manual processes to create a system capable of quickly identifying relationships and providing actionable intelligence to law enforcement.

According to the Trafficking in Persons Report, released by the U.S. Department of State in June 2020, the anti-trafficking hotline received 136,990 calls, texts, chats, online tips and emails during 2019, leading to the identification of 11,852 potential human trafficking cases in the United States alone. That allowed U.S. workers to provide resources and referrals to 3,828 potential victims and identify 3,599 potential cases.

Even with all of those potential traffickers and victims identified, only 1,086 investigations had enough information to go forward. And that’s likely only a fraction of the total breadth of the problem, since the nature of human trafficking itself is obfuscated and shrouded in secrecy. Victims don’t feel empowered to come forward and traffickers operate across a myriad of communication mediums.

To help law enforcement make a bigger dent in this, Marinus needed to be sure that Traffic Jam could not just capture a great deal of data from internet resources but also rapidly turn the information into actionable intelligence. Since most victims would not, or could not, contact law enforcement, Traffic Jam would need to be a tool giving law enforcement another layer of proactive and investigative knowledge.

Cara Jones and Emily Kennedy founded Marinus Analytics in 2014 and launched the company’s flagship application in 2015. Traffic Jam was built using research from work Kennedy started with her senior honors thesis at Carnegie Mellon University. It now serves law enforcement agencies at all levels in the U.S., Canada, and the U.K.

Traffic Jam collects data from websites known to be frequented and used by human traffickers and relies on internal research and feedback from users globally to help identify other potential sources and digital operations.

“We created a repository of online classifieds resources where victims are often found, and that typically comes in the structure of a title and a body of text along with imagery, date and location,” said Jones, chief executive of Marinus Analytics.

During 2019, Marinus successfully used relational database technology to identify at least 3,800 trafficking victims, a majority of whom were in the United States, and identified organized crime by detecting patterns in collected data. However, the process of link charting, the process of identifying and categorizing relationships, was manual and time-consuming at the time.

Using its system, Traffic Jam collected over 40,000 points of data each day – much of that with a great deal of external context that links data points together. Although most data points include such information as imagery, dates, locations, names or pseudonyms, and other harvest data, using a relational database to link together and contextualize the data quickly became a limitation at large scale.

At the time, Marinus already ran Traffic Jam on the Amazon Web Services Inc. cloud and used Amazon RDS, a relational database product, which worked well enough to help provide a single-degree data relationships, but Jones and Kennedy wanted to be able to deliver robust links between numerous data points for law enforcement.

Changing the course of Traffic Jam

To do this, Marinus turned to Amazon Neptune, a cloud-based graph database service that allows the rapid development of apps that can use metadata to create links through highly connective datasets. That opened up the opportunity for Marinus to provide deeper analysis of the collected data and reveal relationships that otherwise might be difficult to discover using a simple relational-style database.

Integration with Neptune began in January 2020 and AWS provided ongoing support for Marinus throughout the process. As soon as the new integration came online, Marinus immediately saw a significant impact from the new graph database.

“We can start to bring in degrees of separation to show that ad A is within two degrees of separation of ad B because they share a phone number,” Marinus Analytics engineering lead Ray Giorgi explained. “Or now ad A is within three degrees of separation of ad C because ads A and B share that phone number, and then ads B and C share an email address.”

Once connected through Neptune, Traffic Jam received a giant boost to its capabilities. The system could now traverse relationships between data points up to 2.5 degrees of separation and expand to collect more than 350,000 data points daily.

Neptune also provides considerable speed to the graph matching capabilities, above and beyond standard relational databases, meaning that now queries run more efficiently and run twice the speed despite using a dataset that is now four to eight times larger than before.

The system is also entirely managed and run within AWS, which means that employees of Marinus can spend more time focusing on Traffic Jam and pushing its capabilities, while Amazon manages the underlying Neptune cluster.

Humans + technology

Human trafficking is a “hidden crime” that requires a proactive solution because victims have been coerced and not informed as to their civil rights and civil liberties, Jones told SiliconANGLE in an interview.

“We know it’s true that there’s not the same sort of self-identification as with a violent crime victim so we have to look for these proactively to help the cause,” said Jones. “With knowledge graphs we now have the data to find the worst of the worst, the largest forms of exploitation, the root cause — and this has radically improved from where we were 10 years ago.”

Marinus collaborates with many different organizations and the company exists specifically to monitor adult services websites – and others in a similar role – that operate as the market between providers and demand. There are a number of different professions of varying backgrounds who use the tool and the data to research specific parts of that industry and community, including law enforcement and victim services.

“I almost feel as if we have an obligation now,” Jones said. “We have a lot of data and powerful infrastructure, so it’s now important we bring to light the highest priority need for public safety. That’s what the knowledge graph is for. This is very hard to quantify and visualize and the knowledge graph really assists with that.”

A large part of Marinus’s vision is to help expose the longstanding causes and structures that create this systemic problem and to help dismantle them.

Adjacent to this work, Marinus is bringing to bear solutions based on the same type of infrastructure for social workers who work in family services and other human-related labor. That data model, and the social graph it uses, could become a prototype for other social facing organizations.

Photo: Pixabay