RoleRAG: Enhancing LLM Role-Playing via Graph Guided Retrieval

TL;DR

RoleRAG improves character role-playing in LLMs by using graph-guided retrieval to accurately recall character knowledge and maintain consistency, reducing irrelevant or hallucinated content.

Contribution

This paper introduces RoleRAG, a novel retrieval framework that combines entity disambiguation and boundary-aware retrieval from knowledge graphs to enhance LLM character consistency.

Findings

Improves alignment with character knowledge in role-playing tasks.

Reduces hallucinated and irrelevant responses in LLM outputs.

Enhances performance on role-playing benchmarks.

Abstract

Large Language Models (LLMs) have shown promise in character imitation, enabling immersive and engaging conversations. However, they often generate content that is irrelevant or inconsistent with a character's background. We attribute these failures to: (1) the inability to accurately recall character-specific knowledge due to entity ambiguity, and (2) a lack of awareness of the character's cognitive boundaries. To address these issues, we propose RoleRAG, a retrieval-based framework that integrates efficient entity disambiguation for knowledge indexing with a boundary-aware retriever for extracting contextually appropriate information from a structured knowledge graph. Experiments on role-playing benchmarks show that RoleRAG's calibrated retrieval helps both general-purpose and role-specific LLMs better align with character knowledge and reduce hallucinated responses.