Propaganda-based Classification of Arabic Newspapers

In this project, we explore the factuality of reporting and the bias of Arabic news media. We use a quantitative measure of political leaning and bias of a set of the most popular newspapers in the Middle East and North Africa. Previous studies have focused on classifying the news based on a single factor, for example, fake/true news. In this project, we analyze and characterize most of the well-known Arabic newspapers based on multiple factors of political polarization such as Islamist, liberal, pro-government, independent, … etc. The labels that are assigned to each newspaper have been collected from three reliable sources, namely; (i) Media Bias/Fact Check (MB/FC); (ii) Wikipedia (Arabic); and (iii) Wikipedia English. The labels from the three sources are aggregated to get a reliable and accurate classification of the Arabic newspapers. Our findings will be used to build a media bias tool that allows the user to get the political polarization of any given Arabic newspaper.