Public Service Media: Balancing Values with Recommender Algorithms

Aktivitet: Foredrag og mundtlige bidragKonferenceoplæg


Recommender systems play an increasingly important role for the exposure of media content to users. As one of the traditional goals of recommender systems has been to optimize exposure in an e-commerce context, media theorists and some public service media (PSM) practitioners have expressed concerns regarding how well diversity, social cohesion and agenda-setting goals can or will be reflected if or when recommender systems are implemented at PSM websites: Will also public service media will start producing filter bubbles of like-minded users? On the other hand, many public service media and -broadcasting organizations want to be perceived as relevant, modern and competitive by their users, and thus feel a need to offer media personalization. This paper presents a comparative empirically driven study of seven European public service media organizations in their implementation of recommender systems. The paper examines how organizations balance the user-value proposition of personalized media with the public service media objectives and the programming policies of the organizations. The paper is based on 25 in-depth interviews with project managers, data scientists and programmers directly involved in the implementation of algorithmic recommender services for public service media. The paper offers a first-hand account on how the much-debated and contested concept 'public service media', as well as its core values such as fairness, diversity and agenda-setting are understood and interpreted by those PSM data practitioners that design and curate the future algorithmically managed exposure of PSM content, beyond broadcast scheduling. A central finding is that personalization only plays a minor role in the production of the algorithmic recommendations.
Periode5 dec. 20197 dec. 2019
BegivenhedstitelPrague Media Point
PlaceringPrague, TjekkietVis på kort
Grad af anerkendelseInternational


  • public service media
  • personalization
  • algorithmic recommender systems
  • East Europe