Why is Normalization Necessary for Linear Recommenders?

2025-04-08

AI Summary

Linear autoencoder (LAE)-based recommendation models suffer from two fundamental biases: popularity bias and neighborhood bias. This paper provides the first theoretical analysis revealing how normalization governs both biases, and proposes Data-Adaptive Normalization (DAN)—a model-agnostic framework that decouples and independently controls these biases via data-driven, dynamic bilateral (user- and item-side) scaling. DAN requires no architectural modifications to underlying LAE models and is universally compatible with existing LAE variants. Extensive experiments across six benchmark datasets demonstrate significant improvements: up to +128.57% in long-tail item recommendation performance and up to +12.36% in unbiased evaluation metrics (e.g., Intra-List Diversity, Coverage). The implementation is publicly available.

Technology Category

• Machine Learning: Learning Preferences or Rankings
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Computer Vision: Bias, Fairness & Privacy

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Despite their simplicity, linear autoencoder (LAE)-based models have shown comparable or even better performance with faster inference speed than neural recommender models. However, LAEs face two critical challenges: (i) popularity bias, which tends to recommend popular items, and (ii) neighborhood bias, which overly focuses on capturing local item correlations. To address these issues, this paper first analyzes the effect of two existing normalization methods for LAEs, i.e., random-walk and symmetric normalization. Our theoretical analysis reveals that normalization highly affects the degree of popularity and neighborhood biases among items. Inspired by this analysis, we propose a versatile normalization solution, called Data-Adaptive Normalization (DAN), which flexibly controls the popularity and neighborhood biases by adjusting item- and user-side normalization to align with unique dataset characteristics. Owing to its model-agnostic property, DAN can be easily applied to various LAE-based models. Experimental results show that DAN-equipped LAEs consistently improve existing LAE-based models across six benchmark datasets, with significant gains of up to 128.57% and 12.36% for long-tail items and unbiased evaluations, respectively. Refer to our code in https://github.com/psm1206/DAN.

Problem

• Addresses popularity bias in linear autoencoder recommenders
• Mitigates neighborhood bias in item correlation modeling
• Proposes adaptive normalization for dataset-specific bias control

Innovation

• Proposes Data-Adaptive Normalization (DAN) for LAEs
• Adjusts item- and user-side normalization adaptively
• Improves performance on popularity and neighborhood biases