Emotionally driven fake news in South Africa

Abstract

To build a more inclusive society that better integrates cyberspace and physical space, we must understand the appeal behind misinformation. Misformation focuses on maintaining an exclusive society rather than integrating society. One of the challenges following the era of information explosion is the rapid spread of misinformation in the form of fake news. People’s choices based on misinformation can have dire consequences, especially in smaller developing communities. Therefore, this paper focuses on the emotional tone of fake news in South Africa to better understand its appeal. Introducing an expected emotional score shows that fake news articles contain more overall emotions than non-fake news. Fake news articles are also written with different biases in mind. These biases were detected and separated using clustering algorithms. Introducing a transformer model allowed us to further classify different biases by creating a profile of the emotions each bias contains. It is found that fake news in South Africa contains a roller-coaster of strong emotive words combining feelings of anger, joy, sadness and fear. The ratio of how these words are combined depends on a particular bias. These findings can help build better detectors of fake news in the future and create a feedback loop to help write more captivating news articles to foster a more inclusive society.