RecNet: Self-Evolving Preference Propagation for Agentic Recommender Systems

TL;DR

RecNet tackles sparse and noisy explicit interactions in agentic recommender systems by introducing a proactive, self-evolving preference propagation mechanism. It combines a centralized router-based forward propagation with a personalized reception module and a textual, feedback-driven optimization loop that uses LLMs for credit assignment and module updates. The approach yields improved modeling of dynamic user and item preferences and demonstrates robust gains across multiple datasets and model backbones, while maintaining efficiency through router-mediated dissemination and asynchronous updates. This framework enables scalable, real-time, and personalized recommendations in evolving networks, representing a practical advance for next-generation recommender systems.

Abstract

Agentic recommender systems leverage Large Language Models (LLMs) to model complex user behaviors and support personalized decision-making. However, existing methods primarily model preference changes based on explicit user-item interactions, which are sparse, noisy, and unable to reflect the real-time, mutual influences among users and items. To address these limitations, we propose RecNet, a self-evolving preference propagation framework that proactively propagates real-time preference updates across related users and items. RecNet consists of two complementary phases. In the forward phase, the centralized preference routing mechanism leverages router agents to integrate preference updates and dynamically propagate them to the most relevant agents. To ensure accurate and personalized integration of propagated preferences, we further introduce a personalized preference reception mechanism, which combines a message buffer for temporary caching and an optimizable, rule-based filter memory to guide selective preference assimilation based on past experience and interests. In the backward phase, the feedback-driven propagation optimization mechanism simulates a multi-agent reinforcement learning framework, using LLMs for credit assignment, gradient analysis, and module-level optimization, enabling continuous self-evolution of propagation strategies. Extensive experiments on various scenarios demonstrate the effectiveness of RecNet in modeling preference propagation for recommender systems.

Figures (6)

• Figure 1: An Analogy between Internet Routing Mechanism and RecNet Preference Propagation Mechanism.
• Figure 2: The overall framework of RecNet.
• Figure 3: The operation of module optimizer considering different optimized modules.
• Figure 4: Performance comparison w.r.t. different backbone models on CDs and Office.
• Figure 5: The self-evolving phenomenon of feedback-driven propagation optimization w.r.t. different initial K.