RuleAgent: Discovering Rules for Recommendation Denoising with Autonomous Language Agents

2025-03-30

AI Summary

In real-world recommender systems, implicit feedback (e.g., clicks) is severely corrupted by unintentional interactions—such as accidental taps or curiosity-driven browsing—introducing substantial noise. Conventional rule-based denoising approaches suffer from poor generalizability and limited transferability across domains. To address this, we propose the first language-agent framework for recommendation denoising, integrating dynamic task planning, reflective reasoning, and role-aware configuration. We further introduce LossEraser, a novel *reverse-learning* strategy that eliminates noisy signals without requiring model retraining. Our method automatically discovers interpretable and transferable denoising rules, obviating manual rule engineering. Extensive experiments on multiple benchmark datasets demonstrate significant improvements over state-of-the-art methods—enhancing both recommendation accuracy and data-cleaning efficiency. This work establishes a new paradigm for modeling implicit feedback under realistic, noisy conditions.

Technology Category

• Data Mining & Knowledge Management: Recommender Systems
• Machine Learning: Learning Preferences or Rankings
• Search and Optimization: Learning to Search

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

The implicit feedback (e.g., clicks) in real-world recommender systems is often prone to severe noise caused by unintentional interactions, such as misclicks or curiosity-driven behavior. A common approach to denoising this feedback is manually crafting rules based on observations of training loss patterns. However, this approach is labor-intensive and the resulting rules often lack generalization across diverse scenarios. To overcome these limitations, we introduce RuleAgent, a language agent based framework which mimics real-world data experts to autonomously discover rules for recommendation denoising. Unlike the high-cost process of manual rule mining, RuleAgent offers rapid and dynamic rule discovery, ensuring adaptability to evolving data and varying scenarios. To achieve this, RuleAgent is equipped with tailored profile, memory, planning, and action modules and leverages reflection mechanisms to enhance its reasoning capabilities for rule discovery. Furthermore, to avoid the frequent retraining in rule discovery, we propose LossEraser-an unlearning strategy that streamlines training without compromising denoising performance. Experiments on benchmark datasets demonstrate that, compared with existing denoising methods, RuleAgent not only derives the optimal recommendation performance but also produces generalizable denoising rules, assisting researchers in efficient data cleaning.

Problem

• Denoising implicit feedback in recommender systems
• Autonomous discovery of generalizable denoising rules
• Reducing manual effort and frequent retraining costs

Innovation

• Autonomous language agents for rule discovery
• Tailored modules enhance reasoning capabilities
• LossEraser unlearning strategy avoids retraining