Test Time Embedding Normalization for Popularity Bias Mitigation

Abstract

Popularity bias is a widespread problem in the field of recommender systems, where popular items tend to dominate recommendation results. In this work, we propose 'Test Time Embedding Normalization' as a simple yet effective strategy for mitigating popularity bias, which surpasses the performance of the previous mitigation approaches by a significant margin. Our approach utilizes the normalized item embedding during the inference stage to control the influence of embedding magnitude, which is highly correlated with item popularity. Through extensive experiments, we show that our method combined with the sampled softmax loss effectively reduces popularity bias compare to previous approaches for bias mitigation. We further investigate the relationship between user and item embeddings and find that the angular similarity between embeddings distinguishes preferable and non-preferable items regardless of their popularity. The analysis explains the mechanism behind the success of our approach in eliminating the impact of popularity bias. Our code is available at https://github.com/ml-postech/TTEN.

arXiv comment5 pages, CIKM 2023

References in corpus (7)

• Adam: A Method for Stochastic Optimization
• BPR: Bayesian Personalized Ranking from Implicit Feedback
• Self-supervised Graph Learning for Recommendation
• LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation
• Model-Agnostic Counterfactual Reasoning for Eliminating Popularity Bias in Recommender System
• The Unfairness of Popularity Bias in Recommendation
• SimpleX: A Simple and Strong Baseline for Collaborative Filtering

Cited by in corpus (2)

• A Survey on Popularity Bias in Recommender Systems
• On Inherited Popularity Bias in Cold-Start Item Recommendation