The New Machine Learning Algorithm Classifies Bullies On Twitter

With 90% accuracy, the newly-developed algorithm is able to identify aggressors and bullies available on the famous social networking service Twitter.

Researchers have recently published a study in the Transactions on the Web (TWEB) that focused on analyzing the behavioral patterns concerning offensive accounts along with their differences from others.

According to Jeremy Blackburn, a study researcher at Binghamton University, US, "We built crawlers -- programs that collect data from Twitter via variety of mechanisms. We gathered tweets of Twitter users, their profiles, as well as (social) network-related things, like who they follow and who follows them."

Then, they conducted sentiment analysis and natural language processing on many tweets. They also analyzed the social network to find out the links between its users.

Their algorithms were developed to classify automatically two particular types of abusive online behavior such as cyber-aggression and cyber-bullying.

The new algorithms can pinpoint abusive Twitter accounts who perform harassing behaviors such as making racist remarks or sending death threats. The accuracy of the new machine learning initiative can reach up to 90%.

Blackburn added, "In a nutshell, the algorithms 'learn' how to tell the difference between bullies and typical users by weighing certain features as they are shown more examples."

He further said, "Our research indicates that machine learning can be used to automatically detect users that are cyber-bullies, and thus could help Twitter and other social media platforms remove problematic users."

Cyber-bullying has been one of the major problems that several social media platforms need to deal with to protect the personal rights of users. Since Twitter always belongs to the list of the frequently used social networking sites, offensive behaviors such as bullying and aggression should be removed as soon as possible.