Degenerate Feedback Loops in Recommender Systems

Abstract

Machine learning is used extensively in recommender systems deployed in products. The decisions made by these systems can influence user beliefs and preferences which in turn affect the feedback the learning system receives - thus creating a feedback loop. This phenomenon can give rise to the so-called "echo chambers" or "filter bubbles" that have user and societal implications. In this paper, we provide a novel theoretical analysis that examines both the role of user dynamics and the behavior of recommender systems, disentangling the echo chamber from the filter bubble effect. In addition, we offer practical solutions to slow down system degeneracy. Our study contributes toward understanding and developing solutions to commonly cited issues in the complex temporal scenario, an area that is still largely unexplored.

Cited by in corpus (27)

• Model-Agnostic Counterfactual Reasoning for Eliminating Popularity Bias in Recommender System
• Measuring the Business Value of Recommender Systems
• Harms from Increasingly Agentic Algorithmic Systems
• Understanding Echo Chambers in E-commerce Recommender Systems
• Machine Culture
• Building Human Values into Recommender Systems: An Interdisciplinary Synthesis
• Auditing YouTube's Recommendation Algorithm for Misinformation Filter Bubbles
• On the User Behavior Leakage from Recommender System Exposure
• AI and the Problem of Knowledge Collapse
• Beyond Ads: Sequential Decision-Making Algorithms in Law and Public Policy
• Recommender Systems and Algorithmic Hate
• User Tampering in Reinforcement Learning Recommender Systems
• A Hippocratic Oath for mathematicians? Mapping the landscape of ethics in mathematics
• Dynamic Causal Collaborative Filtering
• DC-Check: A Data-Centric AI checklist to guide the development of reliable machine learning systems
• Towards Psychologically-Grounded Dynamic Preference Models
• Debiasing Sequential Recommenders through Distributionally Robust Optimization over System Exposure
• Treatment Effect Estimation for User Interest Exploration on Recommender Systems
• A Value-Centered Exploration of Data Privacy and Personalized Privacy Assistants
• Reconciling the Quality vs Popularity Dichotomy in Online Cultural Markets
• Incentives and co-evolution: Steering linear dynamical systems with noncooperative agents
• Understanding or Manipulation: Rethinking Online Performance Gains of Modern Recommender Systems
• Performative Debias with Fair-exposure Optimization Driven by Strategic Agents in Recommender Systems
• Existence conditions for hidden feedback loops in online recommender systems
• Is Downloading this App Consistent with my Values? Conceptualizing a Value-Centered Privacy Assistant
• A Non-Parametric Choice Model That Learns How Users Choose Between Recommended Options
• Stability and Efficiency of Personalised Cultural Markets