Recommenders with a mission

News recommenders have the potential to help users filter the enormous amount of news that is available online, and as such may play an important role in determining what information users do and do not get to see. However, current approaches to evaluating recommender systems are often focused on measuring an increase in user clicks and short-term engagement, rather than measuring the user’s and society’s long-term interest in diverse and important recommendations. In this project, we aim to bridge the gap between so-called normative notions of news diversity, as it is known in social sciences and specifically democratic theory, and quantitative metrics necessary for evaluating the recommender system. We construct new evaluation metrics for news diversity and implement those in an open-source library, available on Github.

In this project, we collaborate with a number of media companies, including but not limited to RTL Netherlands, ZDF and NPO. Our work is made possible by the SIDN / Adessium Foundation