Variational Bayesian Personalized Ranking

2025-03-14

AI Summary

Addressing three key challenges in implicit-feedback recommendation—unobservable user preferences, noise interference (false positives/negatives), and popularity bias—this paper proposes the Variational Bayesian Personalized Ranking (VBPR) framework. We derive an optimizable approximate inference objective via ELBO-KL decomposition and introduce, for the first time, an attention-driven latent interest prototype contrastive mechanism. This mechanism enables hard negative mining in the implicit embedding space, simultaneously mitigating noise effects and alleviating long-tail bias. Coupled with feature-distribution uniformity regularization, it effectively disentangles users’ true preferences from behavioral biases. Extensive experiments demonstrate consistent improvements across multiple state-of-the-art backbone models, achieving an average +2.1% gain in NDCG@10. The code and datasets are publicly released.

Technology Category

• Machine Learning: Learning Preferences or Rankings
• Knowledge Representation and Reasoning: Preferences
• Data Mining & Knowledge Management: Recommender Systems

Application Category

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

Abstract

Recommendation systems have found extensive applications across diverse domains. However, the training data available typically comprises implicit feedback, manifested as user clicks and purchase behaviors, rather than explicit declarations of user preferences. This type of training data presents three main challenges for accurate ranking prediction: First, the unobservable nature of user preferences makes likelihood function modeling inherently difficult. Second, the resulting false positives (FP) and false negatives (FN) introduce noise into the learning process, disrupting parameter learning. Third, data bias arises as observed interactions tend to concentrate on a few popular items, exacerbating the feedback loop of popularity bias. To address these issues, we propose Variational BPR, a novel and easily implementable learning objective that integrates key components for enhancing collaborative filtering: likelihood optimization, noise reduction, and popularity debiasing. Our approach involves decomposing the pairwise loss under the ELBO-KL framework and deriving its variational lower bound to establish a manageable learning objective for approximate inference. Within this bound, we introduce an attention-based latent interest prototype contrastive mechanism, replacing instance-level contrastive learning, to effectively reduce noise from problematic samples. The process of deriving interest prototypes implicitly incorporates a flexible hard sample mining strategy, capable of simultaneously identifying hard positive and hard negative samples. Furthermore, we demonstrate that this hard sample mining strategy promotes feature distribution uniformity, thereby alleviating popularity bias. Empirically, we demonstrate the effectiveness of Variational BPR on popular backbone recommendation models. The code and data are available at: https://github.com/liubin06/VariationalBPR

Problem

• Addresses challenges in ranking prediction from implicit feedback data.
• Reduces noise and data bias in collaborative filtering systems.
• Introduces a novel learning objective for popularity debiasing and noise reduction.

Innovation

• Variational BPR integrates likelihood optimization, noise reduction, debiasing.
• Attention-based latent interest prototype contrastive mechanism reduces noise.
• Hard sample mining strategy alleviates popularity bias effectively.