mlsanityData to 14 Oct 2026

Know You Before You Speak: User-State Modeling for LLM Personalization in Multi-Turn Conversation

arXiv 2605.24647 ↗First posted 23 May 2026cs.CLcited by 3

Not on the map. Why?

Abstract (arXiv)

Personalized dialogue requires more than recalling explicit user histories: systems also need to infer hidden user states that evolve through interaction and shape appropriate response strategies. Existing memory- and profile-based methods primarily reuse observable user information, offering limited support for modeling user-state dynamics or selecting actions based on how they shape future user states. We propose PUMA (Prospective User-state Modeling for Action selection), a framework grounded in the Free Energy Principle (FEP) that formulates personalization as decision-making under partial observability, centered on an explicit user state model that captures latent user states and their action-conditioned dynamics. At each turn, PUMA maintains a belief over the user's hidden state, refines the user state model for observation generation and action-conditioned state transition, and selects dialogue actions by minimizing expected free energy, balancing epistemic and pragmatic objectives under a unified criterion. This formulation shifts personalization from passive memory retrieval to model-based decision-making over user evolution. We instantiate PUMA on healthcare-oriented counseling and motivational interviewing benchmarks with latent state annotations for rigorous evaluation. Experiments show that PUMA improves long-horizon dialogue outcomes while maintaining strong response quality, and a cross-dataset study demonstrates more reliable user-state estimation and next-state prediction.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 39 of its 57 references that are in the map. First posted on arXiv 23 May 2026.

Builds on

39 of 57 references.

References · 39

  1. 2024The Llama 3 Herd of Modelscited by 14,715
  2. 2025Qwen3 Technical Reportcited by 8,820
  3. 2023MemoryBank: Enhancing Large Language Models with Long-Term Memorycited by 556
  4. 2024RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrievalcited by 532
  5. 2017A Survey on Dialogue Systems: Recent Advances and New Frontierscited by 162
  6. 2026MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agentscited by 142
  7. 2024Hello Again! LLM-powered Personalized Agent for Long-term Dialoguecited by 123
  8. 2024Aligning to Thousands of Preferences via System Message Generalizationcited by 84
  9. 2024Large Language Models Empowered Personalized Web Agentscited by 66
  10. 2025Reinforced Latent Reasoning for LLM-based Recommendationcited by 47
  11. 2025Measuring What Makes You Unique: Difference-Aware User Modeling for Enhancing LLM Personalizationcited by 45
  12. 2024Crafting Personalized Agents through Retrieval-Augmented Generation on Editable Memory Graphscited by 38
27 more
  1. 2026HumanLM: Simulating Users with State Alignment Beats Response Imitationcited by 37
  2. 2025NextQuill: Causal Preference Modeling for Enhancing LLM Personalizationcited by 30
  3. 2025Think-While-Generating: On-the-Fly Reasoning for Personalized Long-Form Generationcited by 30
  4. 2023Active Preference Inference using Language Models and Probabilistic Reasoningcited by 29
  5. 2023Enhancing Personalized Dialogue Generation with Contrastive Latent Variables: Combining Sparse and Dense Personacited by 24
  6. 2025Latent Inter-User Difference Modeling for LLM Personalizationcited by 21
  7. 2022A Personalized Dialogue Generator with Implicit User Persona Detectioncited by 19
  8. 2025PRIME: Large Language Model Personalization with Cognitive Dual-Memory and Personalized Thought Processcited by 18
  9. 2025Explicit v.s. Implicit Memory: Exploring Multi-hop Complex Reasoning Over Personalized Informationcited by 17
  10. 2025L-MTP: Leap Multi-Token Prediction Beyond Adjacent Context for Large Language Modelscited by 15
  11. 2025Consistent Client Simulation for Motivational Interviewing-based Counselingcited by 14
  12. 2024RLPF: Reinforcement Learning from Prediction Feedback for User Summarization with LLMscited by 13
  13. 2026NextMem: Towards Latent Factual Memory for LLM-based Agentscited by 12
  14. 2023MIRACLE: Towards Personalized Dialogue Generation with Latent-Space Multiple Personal Attribute Controlcited by 11
  15. 2024Few-shot Dialogue Strategy Learning for Motivational Interviewing via Inductive Reasoningcited by 11
  16. 2025CAMI: A Counselor Agent Supporting Motivational Interviewing through State Inference and Topic Explorationcited by 11
  17. 2025Mem-PAL: Towards Memory-based Personalized Dialogue Assistants for Long-term User-Agent Interactioncited by 11
  18. 2025MSL: Not All Tokens Are What You Need for Tuning LLM as a Recommendercited by 10
  19. 2025MemWeaver: A Hierarchical Memory from Textual Interactive Behaviors for Personalized Generationcited by 9
  20. 2025SteerX: Disentangled Steering for LLM Personalizationcited by 7
  21. 2024MAPLE: A Framework for Active Preference Learning Guided by Large Language Modelscited by 6
  22. 2025Exploring the Impact of Personality Traits on Conversational Recommender Systems: A Simulation with Large Language Modelscited by 5
  23. 2026ODAR: Principled Adaptive Routing for LLM Reasoning via Active Inferencecited by 5
  24. 2024Active Inference and Human--Computer Interactioncited by 4
  25. 2025Reinforced Strategy Optimization for Conversational Recommender Systems via Network-of-Expertscited by 4
  26. 2024An Active Inference Strategy for Prompting Reliable Responses from Large Language Models in Medical Practicecited by 3
  27. 2026UniDetect: LLM-Driven Universal Fraud Detection across Heterogeneous Blockchainscited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026Mind2Dialogue: Training Human-Aware Language Models by Simulating User Mental Statescited by 1
  2. 2026Relational Positioning as a Measurable Risk Object: History-Carried Lock-in and Self-Confabulation in Multi-Turn Human-AI Dialoguecited by 0
  3. 2026Rubric-Aware On-Policy Self-Distillation for LLM Personalizationcited by 0