PrLM: Learning Explicit Reasoning for Personalized RAG via Contrastive Reward Optimization

TL;DR

PrLM introduces a reinforcement learning framework that explicitly trains language models to reason over retrieved user profiles for personalized responses, improving alignment with user preferences and robustness across datasets.

Contribution

The paper presents PrLM, a novel contrastive reward-based reinforcement learning approach for explicit reasoning in personalized RAG, addressing limitations of implicit integration in existing models.

Findings

Outperforms existing personalized RAG methods

Robust across different retrieval settings

Effective without annotated reasoning paths

Abstract

Personalized retrieval-augmented generation (RAG) aims to produce user-tailored responses by incorporating retrieved user profiles alongside the input query. Existing methods primarily focus on improving retrieval and rely on large language models (LLMs) to implicitly integrate the retrieved context with the query. However, such models are often sensitive to retrieval quality and may generate responses that are misaligned with user preferences. To address this limitation, we propose PrLM, a reinforcement learning framework that trains LLMs to explicitly reason over retrieved user profiles. Guided by a contrastively trained personalization reward model, PrLM effectively learns from user responses without requiring annotated reasoning paths. Experiments on three personalized text generation datasets show that PrLM outperforms existing methods and remains robust across varying numbers of retrieved profiles and different retrievers.