Leveraging AI and data analytics to counter misinformation in the MENA: experimenting with data patterns on social media in Arabic content

 

An “infodemic” of misinformation has infected the global online media landscape with increasing virulence over the last five years. As new variants of COVID-19 thrive thanks in part to these same forces, it is critical that we dedicate resources to effectively counter online misinformation. This need is particularly urgent in the Arab world (the Middle East and North Africa, or MENA), where a dearth of independent media has forced citizens to rely on social-media platforms for news. Globally, very few resources have been dedicated to advancing evidence-based research or artificial-intelligence tools to detect and counter the spread of misinformation in the Arabic language.

This research project seeks to evaluate the current landscape of misinformation in Arabic by aggregating data from Arabic fact-checking initiatives into a first-of-its-kind multilateral database. Against this data set, the team will perform analysis using natural language processing techniques to detect broad patterns behind how misinformation affects minority groups, public sentiment towards misinformation and patterns behind its propagation.

This research will provide both a comprehensive overview of the problem and a baseline for further research such as testing classic text classification algorithms against Arabic content for possible contribution to future Arabic-specific misinformation detection prototypes. Through publicly accessible tip lines, the project will solicit further contributions to the database and disseminate information-literacy content. It will make its database open-access to independent Arabic fact-checking organizations and provide them with training to analyze their own data samples using it, amplifying impact and scaling reach in the global Arab community.