AbuSearch team wins award at SAS hackathon for AI project

A UK-based team has secured the 'Data for Good' special award in a global hackathon organised by data and AI technology company SAS, for their efforts to combat online child sexual abuse.

The team, named AbuSearch, consisted of members from SAS Partner Katalyze Data and the Child Rescue Coalition. Their objective was to develop a solution capable of identifying users engaging in sharing or producing harmful content related to children online, as well as identifying those grooming potential victims.

To achieve this, AbuSearch analysed chat and forum data from Project NEMESIS, which automates evidence handling for live-streaming abuse cases. They employed natural language processing, data profiling, visualisation, and network analytics techniques. This strategy enabled them to create a solution that can process large volumes of data to generate alerts, identifying malicious users and uncovering networks among them.

Analysts involved in this project can profile text, examine term maps, apply topic modelling, extract locations from messages, and identify the initial sharers of hyperlinks containing harmful content. This analysis utilises various components of the SAS Viya platform, delivering actionable insights efficiently through a summary dashboard in SAS Visual Analytics.

Sara Di Donna, Sales Director at Katalyze Data, said, "After a chance meeting at SAS Innovate on Tour, we recognised the opportunity for Katalyze Data to work with the Child Rescue Coalition to help identify child abusers online.

"Up to 85 per cent of online offenders are hands-on abusers of children, with as many as one in five girls and one in 20 boys experiencing some form of sexual abuse before the age of 18. This project has the potential to enact real change, with NLP uniquely positioned to generate tangible insights."

The Child Rescue Coalition, a Florida-based non-profit operating worldwide, focuses on providing free software to law enforcement agencies to address child sexual abuse and exploitation threats. Phil Attwood, Director of Impact at the Child Rescue Coalition, expressed his sentiments, stating, "I was really proud to work with the team to advance our abilities to identify threat, risk, and harm from different data sources, in close to real-time. Analysing data from a range of online platforms, we were also able to increase our understanding of high harm threats and links between offenders.

"Artificial intelligence technology will have a huge impact in the future in helping us be more efficient and use finite resources for law enforcement more effectively. We hope it will help progress our mission to help officers identify, arrest and prosecute child predators, in the UK and internationally."

Ashley Beck, Senior Law Enforcement Industry Consultant at SAS UK & Ireland, added, "AbuSearch's entry stood out to our team of judges, showcasing the very best of how technology can be used to tackle important issues. NLP makes it possible for computers to interpret text and speech, measure sentiment and determine which parts are important, all of which has been transformational when working with vast amounts of data."

The SAS Hackathon this year had participation from 1,731 registrants across more than 70 countries, with SAS Partners Microsoft and Intel providing support for the event.