Standing Up to Bullying with Machine Learning

A new study out of the University of Wisconsin, Madison, is using machine learning to fight bullying. According to the study website, "We are a university research team that fights bullying using novel machine learning algorithms. Our interdisciplinary research bridges computer science and psychology in order to study bullying in the modern world. Bullying is a serious national health issue. Previous scientific study of bullying used personal surveys in schools, and suffered from small sample size and low temporal resolution. In contrast, our algorithms discover the participants of a bullying episode, their social roles, and their emotional responses from publicly available social media posts. The scientific data we produce can improve the understanding, intervention, and policy-making on bullying."

It continues, "The figure above shows the daily count of bullying traces (tweets about bullying) on Twitter as identified by our algorithm. The spikes are celebrity events. For example, the spike around September 24, 2011 was due to Lady Gaga dedicating a song to bullying victim Jamey Rodemeyer who committed suicide a few days earlier. There is a weekly (7-day) cycle. Our algorithm only identifies a small fraction of bullying traces -- The actual number is much larger."

Learn more here.

Image: Courtesy University of Wisconsin, Madison