< In Effort To Curb Violence In Chicago, A Professor Mines Social Media

AUDIE CORNISH, HOST:

With the grim milestone of 500 homicides already passed this year in Chicago, police there are grappling with a toxic mix of illegal firearms and gang culture. Add to that mix social media. There are gang-affiliated Facebook pages, Twitter handles, YouTube channels. If you want to see images of a kid getting beat down or worse, you can do that. Or, says Desmond Patton, you can look at it as data to be mined. Patton is trying to create an algorithm that will monitor and identify who might be the next victim or shooter. Now a professor at Columbia University, Patton used to be a social worker in Chicago and has studied gang culture online.

DESMOND PATTON: I got involved in this work because young people were dying based on what they say online. That's horrible.

CORNISH: Patton says gangs use social media to boast, to posture and to threaten each other. When I asked him how a virtual beef escalates into a shooting, he says it often starts with smack talk about a victim.

PATTON: There are a number of things that happen. First, there may be a conversation that unfolds. On Facebook, what you're able to see is someone making direct kind of jokes or making fun of the individual that was killed. And we're also able to see video. So, for example, you may have a video of someone at the funeral or they are expressing their love for someone that was killed, and then someone could make comments based on that video as well. Emojis and hashtags are vastly important. And so you may have someone use prayer hands and things like that in order to express grief. And then people can come behind that and then use more threatening and emojis, perhaps the pistol or a devil face to kind of directly intervene in that grieving post as well.

CORNISH: So how does this change the nature of these disputes? How does it help escalate it?

PATTON: How things have changed are, now, when you're making comments, you're making threatening comments online. You can say these things very quickly, very fast. And then those things are broadcast on these platforms. And we have individuals that you may not be connected with reading and interpreting your thoughts, right? And so oftentimes you may have, you know, comments that were perhaps not intended to be threatening that are interpreted as threatening.

CORNISH: But to detect when that language might flare into violence, you need a lot of data. So with the help of the Chicago YMCA, Patton hired two teens in the community to click through pages, scan status updates and interpret emojis to help researchers build up a dataset.

PATTON: For example, young people would spell the name of a street backwards. And the spelling of that street was emblematic of a known gang territory. But if you weren't from that neighborhood or had an understanding of that, then you would have missed what was actually being communicated. We really try to take in the full picture. So we look at the conversation. We look at the events. We look at the tone of tweets. We look at the actual letters. We look at capitalization. We look at the ways in which punctuation is used.

So we take a really in-depth kind of anthropological approach to understanding what's happening. Once we are able to get that deep understanding, then I pass that information on to our data scientists. And they are able to then develop tools that can detect instances of aggression and grief. And we're able to identify those two themes because they were the most prominent themes in our in-depth look at the language and the context in the tweets.

CORNISH: Slang is always changing.

PATTON: Yes.

CORNISH: You know, there's a lot of parents out there listening (laughter) who know that very well. What are the dangers here? How do you account for that in something like this?

PATTON: Right. Slang changes month to month, year to year. And one of the things that we clearly understood at the beginning of this research process is that we don't know what kids are saying. We have no idea. I have worked in Chicago for many years, and I have worked in many of the neighborhoods that I'm now researching, and I have no understanding of what's being said.

(LAUGHTER)

PATTON: And so I think, you know, having a human-centered approach is always going to be important to doing this type of research. We always need to be - be in partnership with the community on how to understand not only the language, but the context of the language as well. So it's not just words. It's what these words are embedded in, what experience these words are embedded in.

CORNISH: So now comes the point where I have to ask about, like, why open this Pandora's box at all? I mean, we already have, like, civil rights advocates on edge about surveillance. You have other people concerned about racial profiling and these kinds of predictive programs. Can you see why people would be suspicious of this?

PATTON: Absolutely. But I think if we can create tools that are in the right hands and to be able to get young people service and supports that they need in real time, then that can help to prevent deeper connections with the criminal justice system. We really want to figure out ways to get young people the support that they need.

CORNISH: Patton says he wants to develop this for anti-violence programs for other social workers. To him, this is not about funneling kids into the criminal justice system. Before we finished our talk, I asked Patton if there was a case from his research that sticks with him, someone he thinks about as he works with data scientists to get this algorithm up and running. Yeah, he says. Her name was Gakirah Barnes, and she was murdered back in 2014.

PATTON: She was a very well-known gang member on the South Side of Chicago. And one of things that has been said about her is that she was this tough, you know, individual that was willing to kill at the drop of a hat. And she allegedly had up to 20 bodies associated with her by the time she was 17 years old. But when we dig deeper into Gakirah's tweets, she was like any other kid. She was experiencing trauma and grief on a day-to-day basis. And she expressed a lot of pain associated with that grief.

So this is not to, you know, negate the fact that she may have engaged in some very harsh activities, but was a child, and she had childhood experiences that rocked her to her core.

CORNISH: Including several friends - very close friends - who were also shot or killed.

PATTON: Absolutely. And so when these situations happened, she would go to Twitter and say, you know, the pain is unbearable. People don't understand my pain. What if we understood her pain? What if someone would have saw that and said, wow, Gakirah, I - you know, you're going through a lot right now. Let's talk about it. And there have been some instances in various media outlets where people try to communicate with her. But I wonder if they were able to bring in this insight about her deepest, most emotional experiences. If they knew that, if they, instead of looking at her as a gang member, looked at her as a young woman that was experiencing trauma and pain, how would that communication be different? So I just think that social media gives us an opportunity to really dig deeper into these experiences.

CORNISH: Desmond Patton, thank you so much for speaking with us.

PATTON: Thank you so much for having me.

CORNISH: Desmond Patton is an assistant professor at the Columbia School of Social Work.

Copyright © 2016 NPR. All rights reserved. Visit our website terms of use and permissions pages at www.npr.org for further information.

Accuracy and availability of NPR transcripts may vary. Transcript text may be revised to correct errors or match updates to audio. Audio on npr.org may be edited after its original broadcast or publication. The authoritative record of NPR’s programming is the audio record.