The Influence of Media Bias on News Recommender Systems

Media bias has the potential to influence users’ perceptions and disrupt social stability. In the modern media landscape news recommendation algorithms play a significant role in delivering news to users. With this proliferation of online news consumption through recommendation algorithms, and increasing concerns about media bias, understanding the relationship between media bias and news recommendation algorithms is crucial. This research at the SFI Centre for Research Training in Machine Learning examines how media bias affects the news content provided by recommendation algorithms, and how it is consumed by users. We seek to unveil the extent to which these algorithms amplify or mitigate bias in news consumption and their impact on users and society. Our research involves detecting media bias in news articles based on natural language processing techniques and machine learning. We have designed a simulation framework and implemented evaluation experiments to assess the role of recommendation algorithms in propagating media bias. Our results demonstrate that some state-of-the-art news recommendation algorithms are influenced by media bias, resulting in the increased propagation of media bias, particularly among users who show a preference for biased news content. Our work fills a gap in the research concerning the interactions between media bias and news recommender systems. Emphasising the importance of algorithmic transparency, our research is a guideline for designing and monitoring news recommendation algorithms to mitigate their potentially negative impacts. We also call for continued reflection on the ethical and social responsibility of news recommendation platforms in addressing media bias.