Virginia Tech® home

Researchers looking for patterns to combat cyberbullying

Loading player for https://video.vt.edu/media/1_00tftwgn...
Category: research Video duration: Researchers looking for patterns to combat cyberbullying
A group of researchers within University Libraries at Virginia Tech is working on a project that looks to combat cyberbullying. By analyzing and classifying tweets from Twitter, they hope to find patterns that can provide answers and create a positive impact for online communication. 
I think a number of people would agree that bullying, specifically nowadays, cyberbullying, is an issue, and so easy to do. It is so easy to put your thoughts out there and because you're hiding behind a screen, you feel more freedom and say what you want. Dealing with cyberbullying as a major issue in our society is very important and that's why us partaking in this research to try to combat it, or at least help identify it, is pretty meaningful to us. We can also identify if cyberbullying, if a certain time of day is a factor of when cyberbullying occurs also. In the mornings, more people are prone to being cyberbullied because there are more cyberbullies in that hour. Giving those kind of more numeric values, those tweets, those interactions, we can better model how the Twittersphere almost is working. Gaining a better understanding of how people are talking on these platforms and seeing, and how much of all this information that is being consumed is toxic, is civil, is uncivil, is bullying, is not. If you find that a significant portion of part of the space is being uncivil, it's like, okay, there might be some problem going on over here. Let's go look at this more in detail. So I can see a big impact in being able to identify hotspots. So collecting tweets takes a long, long time. Essentially, we would need tens of millions or in general millions of tweets to try to actually come up with a viable answer. How do we choose these people? Where do we start? So we look to see who's following the actual Twitter handle. A lot of people are. So we started looking at followers from people who follow the Twitter handle itself. And then we just set up a harvesting in using Tweepy in Python and gave it a time range and said "go" and just sat there for X number amount of time, and that's how we collected our data. Because they remember stuff and I don't. I like working on a team and these are very capable students. It's fun to kind of guide them from what I've learned in some of the aspirations that I've, I've seen, I have, but also learn from them.