Baylor professors use AI to identify online listings that lead to criminal activity

WACO, Texas (KWTX) – Imagine if a computer could identify a suspicious online listing just by what it says. Thanks to a new grant, two Baylor professors can help make that happen.

Assistant Professors of Computer Science Pablo Rivas and Thomas Carney received $314,284 from the National Science Foundation. They are working with a group of researchers from other universities to prevent sex trafficking and the sale of stolen car parts.

The team is testing AI to determine how sites that sell goods, such as Craigslist, can lead to such illegal activities.

“The group is trying to understand how people communicate on these sites, especially criminal organizations,” Rivas said.

Specifically, the team investigates the words in the listings of these sites and how they can lead to sex trafficking or the sale of stolen car parts. One day, they hope to equip law enforcement with the technology.

“When you get a new leader, suddenly you’re putting together pieces of the puzzle that didn’t make sense before, but suddenly you have multiple perspectives to look at a problem,” Carney said.

Natural language processing (NLP) is a branch of computer science that helps computers understand words better than humans do. It’s not a new technology, but researchers say it’s never been used this way before.

“Now we can take this on an unprecedented scale to provide a better understanding of how language works,” Rivas said.

Imagine AI being able to identify something suspicious from misspelled words or even emojis used in a post. This is what technology can do.

“We’re hoping to look beyond those nuances and look at the potential of a post that’s seen as human trafficking or a post that says it’s an illegal part of a car,” Rivas said.

Undergraduates, graduate students, and doctoral students are all involved in helping Baylor professors with this project. Through the grant, the research project will continue until April 2024

Copyright 2022 KWTX. All rights reserved.