User-Creator Feature Polarization in Recommender Systems with Dual Influence

TL;DR

This paper models the dual influence of recommender systems on users and creators, proving that such systems tend to polarize and reduce diversity, and explores methods to mitigate this effect.

Contribution

It introduces a new model capturing dual influence in recommender systems and analyzes its impact on polarization and diversity, offering insights into effective mitigation strategies.

Findings

Recommender systems with dual influence inherently cause polarization.

Common diversity-promoting methods are ineffective under dual influence.

Relevancy-focused methods like top-k truncation can prevent polarization.

Abstract

Recommender systems serve the dual purpose of presenting relevant content to users and helping content creators reach their target audience. The dual nature of these systems naturally influences both users and creators: users' preferences are affected by the items they are recommended, while creators may be incentivized to alter their content to attract more users. We define a model, called user-creator feature dynamics, to capture the dual influence of recommender systems. We prove that a recommender system with dual influence is guaranteed to polarize, causing diversity loss in the system. We then investigate, both theoretically and empirically, approaches for mitigating polarization and promoting diversity in recommender systems. Unexpectedly, we find that common diversity-promoting approaches do not work in the presence of dual influence, while relevancy-optimizing methods like top-$k$ truncation can prevent polarization and improve diversity of the system.