Disagreement as Signals: Dual-view Calibration for Sequential Recommendation Denoising

2026-04-27

AI Summary

This work addresses the challenge of user behavior noise interfering with accurate interest modeling in sequential recommendation by proposing DC4SR, a dual-view calibration framework. DC4SR uniquely integrates semantic priors derived from a fine-tuned large language model with dynamic posteriors learned by a Transformer-based recommender. Leveraging the discrepancy between these two views as a signal, the framework iteratively performs collaborative calibration to simultaneously denoise user sequences and capture evolving interests. Extensive experiments demonstrate that DC4SR consistently outperforms existing Transformer-based and LLM-enhanced denoising approaches across various noise settings, achieving superior robustness and recommendation performance.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Sequential Decision Making
• Data Mining & Knowledge Management: Recommender Systems

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Sequential recommendation seeks to model the evolution of user interests by capturing temporal user intent and item-level transition patterns. Transformer-based recommenders demonstrate a strong capacity for learning long-range and interpretable dependencies, yet remain vulnerable to behavioral noise that is misaligned with users' true preferences. Recent large language model (LLM)-based approaches attempt to denoise interaction histories through static semantic editing. Such methods neglect the learning dynamics of recommendation models and fail to account for the evolving nature of user interests. To address this limitation, we propose a Dual-view Calibration framework for Sequential Recommendation denoising (DC4SR). Specifically, we introduce a semantic prior, derived from an LLM fine-tuned via labeled historical interactions, to estimate the noise distribution from a semantic perspective. From the learning perspective, we further employ a model-side posterior that infers the noise distribution based on the model's learning dynamics. The disagreement between the two distributions is then leveraged to jointly refine semantic understanding and learning-aware model-side representations. Through iterative updates, dynamic dual-view calibration is achieved for both the global semantic prior and the model-side posterior, enabling consistent alignment with evolving user interests. Extensive experiments demonstrate that DC4SR consistently outperforms strong Transformer-based recommenders and LLM-based denoising methods, exhibiting enhanced robustness across training stages and noise conditions.

Problem

• sequential recommendation
• behavioral noise
• user interest evolution
• denoising
• learning dynamics

Innovation

• dual-view calibration
• sequential recommendation
• behavioral denoising
• semantic prior
• learning dynamics