Unleashing the Power of Large Language Model for Denoising Recommendation
Shuyao Wang, Zhi Zheng, Yongduo Sui, Hui Xiong
TL;DR
The paper tackles the problem of noisy implicit feedback in recommender systems by introducing LLaRD, a framework that leverages large language models to generate denoising-related knowledge (both user preferences and relational structure) and applies a user-centric chain-of-thought reasoning over interaction graphs. This knowledge is then integrated through a knowledge-enhanced denoising module guided by the Information Bottleneck objective, incorporating a mask-based edge selection and HSIC-based compression to align with recommendation targets. Empirical results on three benchmarks and two backbones show that LLaRD consistently outperforms state-of-the-art denoising methods, with notable gains in robustness to noise and improvements in cold-start scenarios. The work demonstrates a practical, scalable pathway for incorporating LLM-derived semantic and relational insights into denoising and robust recommendations.
Abstract
Recommender systems are crucial for personalizing user experiences but often depend on implicit feedback data, which can be noisy and misleading. Existing denoising studies involve incorporating auxiliary information or learning strategies from interaction data. However, they struggle with the inherent limitations of external knowledge and interaction data, as well as the non-universality of certain predefined assumptions, hindering accurate noise identification. Recently, large language models (LLMs) have gained attention for their extensive world knowledge and reasoning abilities, yet their potential in enhancing denoising in recommendations remains underexplored. In this paper, we introduce LLaRD, a framework leveraging LLMs to improve denoising in recommender systems, thereby boosting overall recommendation performance. Specifically, LLaRD generates denoising-related knowledge by first enriching semantic insights from observational data via LLMs and inferring user-item preference knowledge. It then employs a novel Chain-of-Thought (CoT) technique over user-item interaction graphs to reveal relation knowledge for denoising. Finally, it applies the Information Bottleneck (IB) principle to align LLM-generated denoising knowledge with recommendation targets, filtering out noise and irrelevant LLM knowledge. Empirical results demonstrate LLaRD's effectiveness in enhancing denoising and recommendation accuracy.