daily californian logo

BERKELEY'S NEWS • NOVEMBER 18, 2023

UC Berkeley graduate student fights human trafficking through algorithms

article image

SUPPORT OUR NONPROFIT NEWSROOM

We're an independent student-run newspaper, and need your support to maintain our coverage.

AUGUST 20, 2017

Rebecca Portnoff, a doctoral candidate in the UC Berkeley School of Engineering, developed two algorithms aimed to scan through online sex advertisements and find human trafficking circles.

Portnoff presented her dissertation findings Wednesday at KDD 2017, a data science conference in Canada. The algorithms look through sex advertisements on Backpage, an online classified advertisements site, to find human traffickers, according to Portnoff. There is a difference, she added, between sex advertisements that are consensual and those that are related to human trafficking.

“This idea of being able to group together ads by their true owner — the underlying issue is that we would like to help law enforcement prioritize their focus,” Portnoff said. “They want to focus on people who do not choose and who are being forcibly trafficked.

She worked with four other researchers to write a paper about these algorithms, including professor Damon McCoy at New York University’s Tandon School of Engineering. McCoy said the first algorithm links advertisements to a single writer using stylometry, which is the study of people’s writing styles.

“We can still, just by using the style of text, be able to tell the difference,” Portnoff said. “Say these ads are connected because they’re written in the same style — research has shown it can be particular to a specific individual.”

The second algorithm allowed Portnoff’s research team to link Backpage advertisements to bitcoin transactions, according to McCoy. Through this method, the team was able to find instances where multiple advertisements were purchased by the same Bitcoin account, McCoy said.

The team ran the algorithms over a small set of Backpage advertisements and found a large collection of advertisements, that look “suspicious” according to McCoy. He said they found one bitcoin wallet that spent a little over $150,000 for over 5,000 sex advertisements posted on Backpage, most of which were in the Bay Area and advertised young Asian women.

That, I think, is suspicious,” McCoy said.

Portnoff said she hopes that the two methods will help law enforcement and nonprofits determine which advertisements are linked to human trafficking circles. When she began her doctorate five years ago, she said she already knew wanted to conduct research regarding human trafficking.

For the first two years of her doctorate, Portnoff said she had many conversations with various human trafficking nongovernmental organizations to determine what they needed. Thorn is one nonprofit they work with that focuses on using technology innovation to prevent sexual exploitation of children. Thorn has a whole platform that tries to analyze sex advertisements, and the team hopes to integrate their algorithms into this platform, according to McCoy.

“The idea of research is to do practical things” Portnoff said. “My main desire behind all of this work … is I actually want to build something that’s actually useful.”

Malini Ramaiyer is an assistant news editor. Contact her at [email protected] and follow her on Twitter at @malinisramaiyer.
LAST UPDATED

AUGUST 21, 2017


Related Articles

featured article
A multi-campus research collaboration featuring UC Berkeley professors that will explore the intersection of artificial intelligence and human life officially launched this week.
A multi-campus research collaboration featuring UC Berkeley professors that will explore the intersection of artificial intelligence and human life officially launched this week.
featured article
featured article
VeHICaL — or Verified Human Interfaces, Control, and Learning for Semi-Autonomous Systems — is an interdisciplinary project involving UC Berkeley researchers that seeks to develop new and improved ways of building human cyber-physical systems, like self-driving cars and drones.
VeHICaL — or Verified Human Interfaces, Control, and Learning for Semi-Autonomous Systems — is an interdisciplinary project involving UC Berkeley researchers that seeks to develop new and improved ways of building human cyber-physical systems, like self-driving cars and drones.
featured article
The ASUC Senate reconvened for its first regular Wednesday meeting of the semester after passing a resolution over winter break that will permit loans for student organizations, and senators also updated each other about the coming semester.
The ASUC Senate reconvened for its first regular Wednesday meeting of the semester after passing a resolution over winter break that will permit loans for student organizations, and senators also updated each other about the coming semester.