News Recommender Systems and News Diversity, Two of a Kind? A Case Study from a Small Media Market

Abstract

Content recommender systems have become commonplace in all digital platforms, and they profoundly alter the media content presented to users. This also applies to news recommender systems (NRSs) used by media companies. However, as it is generally accepted that diverse news coverage is crucial to maintain democratic societies, the role of NRSs is frequently questioned. We assess the development processes of NRSs at three media companies: two private ones active in several European countries, and one public service broadcaster, using a previously established conceptual model for news diversity. Ultimately, we find that news personalization is currently still more of an idea than an actual recurrent practice, and that the application of NRSs is integrated and harmonized across countries and/or other types of new media types (e.g., streaming services) of the same company. We highlight similarities and differences in approaches, objectives and rationales between private and public companies.

We’re in This Together: A Multi-Stakeholder Approach for News Recommenders

“…In this strategy the use of recommender systems gained particular attention as it provides new opportunities to engage with news audiences (Monzer et al 2020;Newman 2020). This has triggered a widespread academic interest in news recommenders mainly focusing on the diversity of news content that is recommended to users, since news media has been frequently linked to safeguarding democratic principles and systems (Helberger 2019;Hendrickx, Smets, et al 2021;Karppinen 2013;Sup Park 2014). This concern has been popularized by Pariser's (2011) 'filter bubble' hypothesis, assuming that preference-driven algorithms expose users to a poorly diversified news diet.…”

DCAN: Diversified News Recommendation with Coverage-Attentive Networks

“…However, improving the diversity of recommended content is essential for enhancing [1], [48], [37] user experience. The excessive pursuit of accuracy indirectly leads to information redundancy in recommended content [39], users' boredom [23], [25] and filter bubbles [17], [35], [46], which have many negative effects on society.…”