Researchers at a British university have created a new algorithm that uses artificial intelligence to predict which Twitter users are going to spread disinformation before they do it.
The machine-learning algorithm was developed by a team of researchers at the University of Sheffield, led by PhD student Yida Mu and Dr. Nikos Aletras from the university’s Department of Computer Science. It can pinpoint with 79.7% accuracy which users are likely to share content from a news source deemed to be unreliable.
To create the algorithm, the researchers analyzed over 1 million publicly available tweets from approximately 6,200 Twitter aficionados. Users were then split into those who shared unreliably sourced news and those who shared reliably sourced news, and this data was used to train the algorithm.
The study found that Twitter users who shared stories from unreliable sources were more likely to write about the world around them, posting on social media about politics or religion. Words used frequently by this category of users included “liberal,” “government,” “Islam,” “Israel,” and “media.”
Twitter users who shared stories from news sources the study categorized as reliable were more focused on themselves, often tweeting about their emotions and personal lives and favoring the words “I’ll,” “birthday,” “wanna,” and “mood.”
The reliable news sharers were found to express their views in language that was more polite than that used by the sharers of disinformation. Rude language and the spread of unreliable content were found to correlate with high online political hostility.
“Social media has become one of the most popular ways that people access the news, with millions of users turning to platforms such as Twitter and Facebook every day to find out about key events that are happening both at home and around the world,” said Dr. Nikos Aletras, lecturer in Natural Language Processing at the University of Sheffield.
“However, social media has become the primary platform for spreading disinformation, which is having a huge impact on society and can influence people’s judgement of what is happening in the world around them.”
The study, “Identifying Twitter users who repost unreliable news sources with linguistic information,” was published in PeerJ journal.