Mitigating Popularity Bias in Recommendation with Unbalanced Interactions: A Gradient Perspective

TL;DR

This paper introduces a gradient-based framework to mitigate popularity bias in recommender systems, normalizing embeddings and adjusting gradients to improve personalization and reduce over-recommendation of popular items.

Contribution

It offers a novel, model-agnostic approach that normalizes user embeddings and adjusts gradients during testing to effectively reduce popularity bias in recommendations.

Findings

Significantly reduces popularity bias in multiple models.

Improves recommendation accuracy and personalization.

Outperforms state-of-the-art debiasing methods.

Abstract

Recommender systems learn from historical user-item interactions to identify preferred items for target users. These observed interactions are usually unbalanced following a long-tailed distribution. Such long-tailed data lead to popularity bias to recommend popular but not personalized items to users. We present a gradient perspective to understand two negative impacts of popularity bias in recommendation model optimization: (i) the gradient direction of popular item embeddings is closer to that of positive interactions, and (ii) the magnitude of positive gradient for popular items are much greater than that of unpopular items. To address these issues, we propose a simple yet efficient framework to mitigate popularity bias from a gradient perspective. Specifically, we first normalize each user embedding and record accumulated gradients of users and items via popularity bias measures in model training. To address the popularity bias issues, we develop a gradient-based embedding adjustment approach used in model testing. This strategy is generic, model-agnostic, and can be seamlessly integrated into most existing recommender systems. Our extensive experiments on two classic recommendation models and four real-world datasets demonstrate the effectiveness of our method over state-of-the-art debiasing baselines.