RGMem: Renormalization Group-based Memory Evolution for Language Agent User Profile

2025-10-18

AI Summary

Current LLM-based dialogue systems suffer from fixed-context windows and static memory mechanisms, hindering the modeling of long-term user states and behavioral consistency across sessions. Mainstream approaches—such as RAG and explicit memory—support only fact-level storage and lack the capacity to distill deep preferences and latent user characteristics from fragmented, multi-turn dialogues. To address this, we propose a dynamic memory evolution framework inspired by the physics-inspired renormalization group (RG) theory, leveraging multi-scale information compression and emergent phenomena. Our method hierarchically coarse-grains and rescales dialogue histories, jointly optimizing semantic extraction and user insight to construct an explicitly structured, self-evolving memory. Experiments demonstrate significant improvements in user profile depth and robustness: the framework enables persistent long-term memory updates under noise, enhances personalized response quality, and strengthens cross-session coherence.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Natural Language Processing: Conversational AI/Dialog Systems
• Machine Learning: Life-Long and Continual Learning

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Personalized, context-aware and across-device search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Personalized and continuous interactions are the key to enhancing user experience in today's large language model (LLM)-based conversational systems, however, the finite context windows and static parametric memory make it difficult to model the cross-session long-term user states and behavioral consistency. Currently, the existing solutions to this predicament, such as retrieval-augmented generation (RAG) and explicit memory systems, primarily focus on fact-level storage and retrieval, lacking the capability to distill latent preferences and deep traits from the multi-turn dialogues, which limits the long-term and effective user modeling, directly leading to the personalized interactions remaining shallow, and hindering the cross-session continuity. To realize the long-term memory and behavioral consistency for Language Agents in LLM era, we propose a self-evolving memory framework RGMem, inspired by the ideology of classic renormalization group (RG) in physics, this framework enables to organize the dialogue history in multiple scales: it first extracts semantics and user insights from episodic fragments, then through hierarchical coarse-graining and rescaling operations, progressively forms a dynamically-evolved user profile. The core innovation of our work lies in modeling memory evolution as a multi-scale process of information compression and emergence, which accomplishes the high-level and accurate user profiles from noisy and microscopic-level interactions.

Problem

• Modeling cross-session long-term user states in LLMs
• Distilling latent preferences from multi-turn dialogues
• Achieving behavioral consistency in personalized conversational systems

Innovation

• Renormalization group-inspired multi-scale memory organization
• Hierarchical coarse-graining extracts latent user preferences
• Dynamic memory evolution through information compression and emergence