Measuring misinformation trends on social media in South Africa using Machine Learning

Abstract

Misinformation, disinformation, malinformation, and/or fake news have gained attention for good and bad in South Africa, especially since the COVID-19 pandemic. The research-based and non-research-based interventions to tackle misinformation have also been slowly gaining traction, particularly through fact checkers, fake news reporting systems such as those by real411, research on automated systems to detect fake news online using machine learning, sentiment analysis of fake news, tagging of fake news data, and so on. Nevertheless, the spread of misinformation and/or fake news still presents a serious threat and challenge to social media platform owners, citizens, lawmakers, governments, and businesses alike. The awareness, engagement, influence, and impact levels of misinformation on citizens, politicians, journalists, and lawmakers are hypothesized to be relatively low, especially in South Africa. However, no sufficient research has been done in this area to understand engagements, awareness, and reporting of fake news online. This research, therefore, uses open-source intelligence and selected machine learning techniques to analyses publicly collected social media data to monitor and measure the awareness and engagements of fake news in South Africa over 30 days. The research further identifies key drivers of spreading or reporting misinformation online. The research concludes that misinformation engagements on social media in South Africa are active, but only in affluent regions and influenced by mobile device users, who are mostly male. The study recommends further research that may support raising misinformation awareness and positive engagements on social media.