Can algorithms help stop gang violence on social media before it spills into the streets?

“Cyber banging” has been making headlines for years, but researchers developing a tool at Columbia University hope computer analysis of language in social media posts can help people on the ground stop gang violence before it happens.

As social media continues to permeate American culture, more and more violence plays out on platforms like Facebook, Twitter and Snapchat. An incident in Chicago this week where four black suspects tied up, taunted and tortured a disabled white man while making racist comments was streamed on Facebook Live, and drew the ire of people nationwide. Last month, a man and a woman were arrested in Milwaukee after they allegedly beat a man to death, posting video of the beating to Snapchat.

These incidents are sometimes seen by millions of people, as they grab national headlines, but in some communities in cities like Chicago, the relationship between violence and social media is less visible and more complex. In March, a man named Brian Fields was shot several times on the city’s South Side while streaming on Facebook Live.

“More and more of these incidents either originate or escalate from some type of activity that is on a social media platform,” said Chicago police spokesman Anthony Guglielmi at the time, noting that it’s standard for Chicago police to search social media after a shooting, looking for clues from instances of the so-called cyber banging, where gang members taunt one another on social media.

And while police are looking at these online platforms after acts of violence, Desmond Patton, a professor at Columbia University in New York, is part of a team that’s looking at language in social media posts and hoping to detect the violence before it starts.

“What we have is a computational system, or an algorithm, that can detect and predict aggression and loss,” said Patton, noting he and his team are focusing their research on a small Twitter-based data set in Chicago.

The researchers hope the tool will be useful in gang-plagued communities like Chicago’s South Side. For Patton, a social media post isn’t always the end result, like the Facebook Live torture video in Chicago this week. Oftentimes, it can be the inciting incident, with gang members taunting one another in videos or tweets.

“At times, that can be in the form of an aggressive or threatening comment, or an emoji or hash tag that indicates gang activity or violent behavior,” said Patton in an interview with Guns.com.

“What we’ve also learned is that there are oftentimes precipitating events that precede those aggressive comments, and so what we’ve noticed is that … exposure to trauma and stress, grief and loss oftentimes are a part of the conversation and escalate on social media platforms, due to the hyper-visibility and connectivity that happens on social media,” he said.

The team hopes to further develop the tool so it can be used to alert violence interrupters, those that do prevention and intervention work in gang-plagued communities, when online communication looks like it’s heating up.

“They will be given a justification as to why we believe that that content is associated with aggression or loss or a threat, and then based on their level of expertise, which we will rely on heavily, then they can make a decision on how to move on that information,” Patton said.

The team is applying for grants and working to scale up the tool to see if it will work in a larger data set. They’re also working with community-based organizations to figure out the best way to deploy the system in intervention scenarios.

Patton says while social media sometimes helps escalate certain situations, people shouldn’t blame the social media platforms for the violence.

“I think the tools are being used by a user and so the tools are going to capture what’s happening in the environment,” he said. “So is it the tool’s fault that the individual was shot, or is there a problem with violence in our country?”

As for the algorithm, Patton is optimistic that he and his team are headed in the right direction.

“Will it work? We’ll see,” he said.

Latest Reviews

revolver barrel loading graphic

Loading