This article by Quentin Villermet, Jérémie Poiroux, Manuel Moussallam, Thomas Louail and Camille Roth has been selected at RecSys ’21: Fifteenth ACM Conference on Recommender Systems in September 2021 where it received the best paper runner-up award.
It focuses on user consumption diversity in online music streaming, showing that there is no blanket answer to the question of the effect of recommendation : it applies differently to different user types. The paper thus speaks of “filter niches” rather than “filter bubbles”: the influence of recommendation depends on users, their behavior and plausibly their expectations toward algorithmic guidance. This idea could shed light on recent findings and debates whereby algorithmic recommendation is sometimes found to be “bubbly”, sometimes not. Whether recommendation expands or not users’ horizon depends on their appetency for such or such affordance – on their filter niche. Also, algorithmic recommendation is generally compared against organic behavior. Yet human recommendation is another comparison point (e.g., platform playlists and radio programs) which this paper uses, thus proposing a trichotomy: organic, algorithmic and editorial access to content.