Problem
Research questions and friction points this paper is trying to address.
personalized retrieval
conversational memory
user-aware recall
memory retrieval
long-term dialogue
Learning User-Aware Recall: Personalized Retrieval in Long-Term Conversational Memory
🤖 AI Summary
This work addresses the limitation of existing conversational memory systems, which often neglect user-specific relevance during retrieval and rely heavily on query similarity or static ranking strategies, leading to suboptimal recall of personally useful memories. To overcome this, the authors propose the PPRO framework, which first constructs episodic and semantic memory stores and generates explicit user profiles as personalized priors to guide memory ranking. Second, it introduces Group Relative Policy Optimization to train a query rewriter that jointly optimizes retrieval and downstream response quality under a fixed memory store and response model. This study is the first to integrate explicit user profiles into memory retrieval ranking and combines retrieval-oriented query rewriting with a reinforcement learning–based feedback mechanism. Experiments demonstrate that PPRO significantly outperforms various baselines on the LoCoMo and LongMemEval-S benchmarks, with ablation studies confirming the effectiveness of profile-guided ranking and retrieval-optimized rewriting.
📝 Abstract
Long-term conversational agents are expected to remember past interactions, but memory is useful only when the right evidence is recalled for the right user. Existing memory-augmented LLM agents have made progress in building compact memory banks, yet retrieval is still often driven by query-centered similarity or fixed ranking rules, leaving user-conditioned relevance underexplored.To address this gap, we propose Profile-guided Personalized Retrieval Optimization (PPRO), a retrieval-centric framework that makes memory retrieval both user-aware and optimizable.PPRO builds episodic and semantic memory banks from dialogue histories and derives a user profile from accumulated memories.The profile serves as an explicit personalized prior in memory ranking, allowing retrieval to account for stable user attributes, preferences, and relationships.PPRO further trains a query rewriter with Group Relative Policy Optimization, using both evidence retrieval quality and downstream answer quality as feedback while keeping the memory banks and answer model fixed.Experiments on LoCoMo and LongMemEval-S show consistent gains over training-free memory systems and training-based baselines.Ablation studies further show that both profile-guided ranking and retrieval-oriented rewriting contribute substantially to performance, highlighting retrieval optimization as a key factor in personalized long-term memory use.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Personalized Retrieval
User Profile
Memory-Augmented LLM
Query Rewriting
Long-Term Conversational Memory
🔎 Similar Papers
No similar papers found.
