News Recommender Systems: Integrating Supply and Demand Perspectives

Up next in this ECREA 2022 session is my temporary University of Zürich colleague Sina Blassnig, whose focus is on news recommender systems. Such systems are algorithms that provide users with personalised recommendations for news content based on past interactions by them or similar users, overall popularity metrics, and other features.

Such systems are increasingly employed by news organisations internationally, and therefore now also need to be investigated from a political communication perspective. They have implications for political news demand, consumption patterns, and citizens’ information practices. Such systems thus connect both sides of the political information environment, but the research still remains fragmented into the supply and demand perspectives, and further synthesising of these perspectives is required.

What do we know already about these systems and their impact on the supply and demand sides, then? A large-scale literature review of articles published in the past eleven years found a number of key patterns: on the supply side, research mainly explores the factors influencing the implementation of news recommender systems; journalistic attitudes towards such systems and their effects on news practices and role perceptions; and the organisational dynamics around such systems. But adoption across different organisations, political and media systems is underresearched; the same goes for the factors influencing journalistic attitudes, or the impact of these systems on power dynamics, decision-making structures, and hierarchies in the organisations.

On the demand side, user attitudes towards algorithmic technologies, effects on news consumption and information reception, and political implications of such systems have been researched; yet there is little systematic insight into the factors influencing such user attitudes, there is a greater focus on social media and aggregator recommender systems than those used by news organisations themselves, and there is less focus on the impact of such systems on political participation.

What we need in future research, then, is fivefold: first, there is a need for more research that integrates supply and demand perspectives. This might build on simulations and agent-based modelling, for example. Second, there is a need for more theory-informed work, as informed for instance by institutional theory used to explain processes of legitimation, imitation, and homogenisation as these systems are adopted across the industry. Third, there’s also a need for more globally comparative work; and fourth, for more interdisciplinary work – for instance between political communication and computer science to combine normative and technological aspects. And finally, as so many of these systems remain black boxes, there is a substantial need for more collaboration between industry and academia that could also draw on more technology testing or user data to better understand these systems. These steps could help address the fragmentation of the field between the supply and demand perspectives.