Recommenders with a Mission

Abstract

News recommenders help users to find relevant online content and have the potential to fulfill a crucial role in a democratic society, directing the scarce attention of citizens towards the information that is most important to them. Simultaneously, recent concerns about so-called filter bubbles, misinformation and selective exposure are symptomatic of the disruptive potential of these digital news recommenders. Recommender systems can make or break filter bubbles, and as such can be instrumental in creating either a more closed or a more open internet. 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 longer term interest in diverse and important information.This paper aims to bridge the gap between normative notions of diversity, rooted in democratic theory, and quantitative metrics necessary for evaluating the recommender system. We propose a set of metrics grounded in social science interpretations of diversity and suggest ways for practical implementations.

Filter bubbles in recommender systems: Fact or fallacy—A systematic review

“…It can be seen as an optimization problem that aims to maximize the diversity metric. • Diversity in items and/or source: It is important to decide whether diversity should be introduced only in the recommended content or in the content provider as well (Vrijenhoek et al, 2021). For example, in online shopping, diversified items may include different garments, while diversified sources may involve different brands.…”

“…It has been observed that this simple approach works well in certain scenarios. It can be seen as an optimization problem that aims to maximize the diversity metric. Diversity in items and/or source : It is important to decide whether diversity should be introduced only in the recommended content or in the content provider as well (Vrijenhoek et al, 2021). For example, in online shopping, diversified items may include different garments, while diversified sources may involve different brands. Personalized/user ‐ specific diversity : Diversity can be introduced irrespective of user profiles, which is referred to as nonpersonalized diversity.…”

Towards a Normative Perspective on Journalistic AI: Embracing the Messy Reality of Normative Ideals

“…The discussion around diversity-sensitive recommender design is a good example to demonstrate why it is so important to acknowledge and embrace the messy reality of normative ideals. In response to the concerns about filter bubbles and echo chambers that have framed much of the scholarly discourse about the societal impact of AI-powered recommender algorithms over the past decade, a growing body of scholarship has started to investigate the potential of translating media diversity as a democratic goal and normative ideal into recommender design (Vrijenhoek et al 2021;Helberger, Karppinen, and D'Acunto 2018;Bernstein et al 2021). This body of literature forms a counterweight to the computer science literature that often approaches diversity as a mathematical problem (Kunaver and Po zrl 2017; see also the excellent overview in Loecherbach et al 2020).…”

Avenues to News and Diverse News Exposure Online: Comparing Direct Navigation, Social Media, News Aggregators, Search Queries, and Article Hyperlinks

“…Turning to other avenues, search and aggregators may limit political diversity due to ranking and recommendation algorithms, which personalize information, recommendations, and search output in ways that align with individuals’ interests, attitudes, and past behavior (Pariser, 2011; Vrijenhoek et al, 2020). At the same time, the search may lead to dissimilar exposure because users see search engines as fair and unbiased (Fallows, 2005; Dutton et al, 2017; Logg et al, 2019) and use rank order, not partisan leaning, as the dominant criterion in selecting news (Nechushtai and Lewis, 2019; Puschmann, 2019).…”