Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MR.Rec: Synergizing Memory and Reasoning for Personalized Recommendation Assistant with LLMs

Jiani Huang, Xingchen Zou, Lianghao Xia, Qing Li

TL;DR

MR.Rec presents a memory-synergized, reasoning-enabled framework for LLM-based personalized recommendations. It deploys a hierarchical memory indexing system (local memory and global memory) and introduces reasoning-enhanced memory retrieval that interleaves memory exploration with multi-step reasoning, all trained via reinforcement learning with verifiable rewards. The approach demonstrates consistent improvements over baselines across multiple datasets, memory settings, and LLM backbones, while delivering efficient indexing and retrieval. Overall, MR.Rec advances interactive, context-aware recommendation assistants by tightly coupling external memory with adaptive reasoning.

Abstract

The application of Large Language Models (LLMs) in recommender systems faces key challenges in delivering deep personalization and intelligent reasoning, especially for interactive scenarios. Current methods are often constrained by limited context windows and single-turn reasoning, hindering their ability to capture dynamic user preferences and proactively reason over recommendation contexts. To address these limitations, we propose MR.Rec, a novel framework that synergizes memory and reasoning for LLM-based recommendations. To achieve personalization, we develop a comprehensive Retrieval-Augmented Generation (RAG) system that efficiently indexes and retrieves relevant external memory to enhance LLM personalization capabilities. Furthermore, to enable the synergy between memory and reasoning, our RAG system goes beyond conventional query-based retrieval by integrating reasoning enhanced memory retrieval. Finally, we design a reinforcement learning framework that trains the LLM to autonomously learn effective strategies for both memory utilization and reasoning refinement. By combining dynamic memory retrieval with adaptive reasoning, this approach ensures more accurate, context-aware, and highly personalized recommendations. Extensive experiments demonstrate that MR.Rec significantly outperforms state-of-the-art baselines across multiple metrics, validating its efficacy in delivering intelligent and personalized recommendations. We will release code and data upon paper notification.

MR.Rec: Synergizing Memory and Reasoning for Personalized Recommendation Assistant with LLMs

TL;DR

MR.Rec presents a memory-synergized, reasoning-enabled framework for LLM-based personalized recommendations. It deploys a hierarchical memory indexing system (local memory and global memory) and introduces reasoning-enhanced memory retrieval that interleaves memory exploration with multi-step reasoning, all trained via reinforcement learning with verifiable rewards. The approach demonstrates consistent improvements over baselines across multiple datasets, memory settings, and LLM backbones, while delivering efficient indexing and retrieval. Overall, MR.Rec advances interactive, context-aware recommendation assistants by tightly coupling external memory with adaptive reasoning.

Abstract

The application of Large Language Models (LLMs) in recommender systems faces key challenges in delivering deep personalization and intelligent reasoning, especially for interactive scenarios. Current methods are often constrained by limited context windows and single-turn reasoning, hindering their ability to capture dynamic user preferences and proactively reason over recommendation contexts. To address these limitations, we propose MR.Rec, a novel framework that synergizes memory and reasoning for LLM-based recommendations. To achieve personalization, we develop a comprehensive Retrieval-Augmented Generation (RAG) system that efficiently indexes and retrieves relevant external memory to enhance LLM personalization capabilities. Furthermore, to enable the synergy between memory and reasoning, our RAG system goes beyond conventional query-based retrieval by integrating reasoning enhanced memory retrieval. Finally, we design a reinforcement learning framework that trains the LLM to autonomously learn effective strategies for both memory utilization and reasoning refinement. By combining dynamic memory retrieval with adaptive reasoning, this approach ensures more accurate, context-aware, and highly personalized recommendations. Extensive experiments demonstrate that MR.Rec significantly outperforms state-of-the-art baselines across multiple metrics, validating its efficacy in delivering intelligent and personalized recommendations. We will release code and data upon paper notification.
Paper Structure (42 sections, 9 equations, 7 figures, 5 tables)

This paper contains 42 sections, 9 equations, 7 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Preliminary
  3. Methodology
  4. Memory Indexing for RAG Recommender
  5. User-specific Local Memory
  6. Cross-User Global Memory
  7. Unifying Memory and Reasoning for Recommendation
  8. Reasoning-enhanced Memory Retrieval
  9. Retrieval-Augmented Item Generation
  10. Enhancing Reasoning with Memory Retrieval
  11. Reinforcement Learning for Memory-synergized Reasoning
  12. Experiments
  13. Experimental Setup
  14. Datasets
  15. Baselines
  16. ...and 27 more sections

Figures (7)

  • Figure 1: Illustration of Personalized and Intelligent LLM-based Recommendation Assistant.
  • Figure 2: The overall model architecture of the proposed MR.Rec framework.
  • Figure 3: Ablation of different components of MR.Rec.
  • Figure 4: Reward trajectories of MR.Rec using the base and instruct models of Qwen-2.5-3B, during our RL tuning.
  • Figure 5: Effect of using different retrievers for MR.Rec.
  • ...and 2 more figures