Researchers at University of Wisconsin in Madison developed a method for analyzing Twitter messages to find tweets, as Twitter messages are called, with evidence of childhood bullying. Educational psychologist Amy Bellmore (pictured left) and colleagues from Wisoconsin’s psychology and computer science departments presented their findings at a North American chapter meeting of the Association for Computational Linguistics earlier this summer.
Researching the problem of school age bullying is difficult, which normally relies on reports from the children themselves, often on self-reporting surveys. “For a standard study we may get access to students from one grade in one school,” says Bellmore. “And then we get a one-time shot at it. We get one data collection point in a school year from these kids. It’s very labor- and time-intensive.”
Bellmore enlisted the help of Wisconsin computer scientist Jerry Zhu, who studies machine learning, to teach computers to scour the voluminous posts on Twitter for mentions of bullying events. Zhu devised a machine-learning algorithm, from reading two sets of tweets: one set about bullying and another set on completely different subjects.
“In machine learning,” says Zhu, “the algorithm reads each tweet as a short text document, and it goes about analyzing the word usage to find the important words that mark bullying events.”
Zhu’s trained system then analyzed some 250 million publicly visible messages posted on Twitter on a daily basis, identifying more than 15,000 bullying-related tweets per day. They found more evidence of bullying appears on weekdays, when children are in school, than on weekends.
The system also identified in tweets the roles of individuals often found in bullying incidents: bullies, victims, accusers, and defenders. But they also found a new role, that of the reporter. “It’s just like it sounds,” says Bellmore, “a child who witnessed or found out about, but wasn’t participating in, a bullying encounter.”
The researchers plan to extend their analysis to other social networks, such as Facebook. Using the data to show the bullied that they are not alone could also help children deal with their feelings. The insights from the research can also help the researchers supply policy-makers with a better understanding of bullying issues, Bellmore and Zhu say, which may result in more effective prevention methods.
Read more: Grants Awarded for Robotics, Cyber-Vision and Child Health
* * *
You must be logged in to post a comment.