engleski

Media bias detection in reporting on the COVID-19 pandemic: The case of Croatian news portals

In extraordinary circumstances accompanied by a high degree of uncertainty, such as the events associated with the COVID-19 pandemic, the objectivity of media coverage becomes even more important. In such situations, people primarily rely on widely available sources whose content can be updated quickly. Therefore, even a subtle deviation from impartial language can act suggestively and contribute to an increase in anxiety levels among the population or to the spread of misinformation. The aim of this paper is to analyse the reporting of Croatian web portals on the COVID-19 pandemic. To this purpose, we collected articles published between mid-March and early July 2020. We annotated the collected data and analysed it using lexical profiling tools. The statistical analysis and comparison of the data reveal differences in word choices, collocational tendencies and sentence complexity and thus bring to light variations in Croatian media reporting strategies. Finally, we employ deep neural networks to further evaluate semantic similarity of various article elements, in order to detect and point out differences that may indicate the presence of bias.

media bias, COVID-19, text analysis, deep neural networks

nije evidentirano