From Atom to Community: Structured and Evolving Agent Memory for User Behavior Modeling

2026-01-23

AI Summary

This work addresses the limitations of current large language models in user behavior modeling, which struggle to capture implicit preferences from non-textual interactions and rely on a single unstructured memory summary, often leading to conflation of multidimensional interests, forgetting of evolving preferences, and data sparsity issues. To overcome these challenges, the authors propose STEAM, a novel framework featuring an atomic and community-based memory architecture. It decomposes user preferences into atomic memory units and leverages cross-user memory communities to propagate collaborative signals. An adaptive memory consolidation and formation mechanism dynamically evolves the memory structure over time. Experiments on three real-world datasets demonstrate that STEAM significantly outperforms state-of-the-art baselines, achieving notable improvements in recommendation accuracy, simulation fidelity, and diversity.

Technology Category

• Machine Learning: Learning Preferences or Rankings
• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

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

Abstract

User behavior modeling lies at the heart of personalized applications like recommender systems. With LLM-based agents, user preference representation has evolved from latent embeddings to semantic memory. While existing memory mechanisms show promise in textual dialogues, modeling non-textual behaviors remains challenging, as preferences must be inferred from implicit signals like clicks without ground truth supervision. Current approaches rely on a single unstructured summary, updated through simple overwriting. However, this is suboptimal: users exhibit multi-faceted interests that get conflated, preferences evolve yet naive overwriting causes forgetting, and sparse individual interactions necessitate collaborative signals. We present STEAM (\textit{\textbf{ST}ructured and \textbf{E}volving \textbf{A}gent \textbf{M}emory}), a novel framework that reimagines how agent memory is organized and updated. STEAM decomposes preferences into atomic memory units, each capturing a distinct interest dimension with explicit links to observed behaviors. To exploit collaborative patterns, STEAM organizes similar memories across users into communities and generates prototype memories for signal propagation. The framework further incorporates adaptive evolution mechanisms, including consolidation for refining memories and formation for capturing emerging interests. Experiments on three real-world datasets demonstrate that STEAM substantially outperforms state-of-the-art baselines in recommendation accuracy, simulation fidelity, and diversity.

Problem

• user behavior modeling
• non-textual behaviors
• structured memory
• preference evolution
• collaborative signals

Innovation

• structured memory
• evolving preferences
• memory communities
• atomic memory units
• collaborative signal propagation