Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

PersonaAgent with GraphRAG: Community-Aware Knowledge Graphs for Personalized LLM

Siqi Liang, Yudi Zhang, Yue Guo

TL;DR

The paper tackles the challenge of personalized AI agents by integrating dynamic user personas with community-aware knowledge via GraphRAG. It introduces a Knowledge Graph that encodes individual interactions and community patterns, and a GraphRAG retrieval mechanism that gathers both user-specific and global context to form personalized prompts for LLMs. Empirical results on the LaMP benchmark show consistent improvements across news categorization, movie tagging, and product rating, including notable gains for smaller models. The work demonstrates the value of combining structured memory with graph-based retrieval to achieve grounded, explainable, and adaptable personalization in LLM-driven agents, with future plans toward multi-agent collaboration and IRL-based preference inference.

Abstract

We propose a novel framework for persona-based language model system, motivated by the need for personalized AI agents that adapt to individual user preferences. In our approach, the agent embodies the user's "persona" (e.g. user profile or taste) and is powered by a large language model (LLM). To enable the agent to leverage rich contextual information, we introduce a Knowledge-Graph-enhanced Retrieval-Augmented Generation (Graph RAG) mechanism that constructs an LLM-derived graph index of relevant documents and summarizes communities of related information. Our framework generates personalized prompts by combining: (1) a summary of the user's historical behaviors and preferences extracted from the knowledge graph, and (2) relevant global interaction patterns identified through graph-based community detection. This dynamic prompt engineering approach allows the agent to maintain consistent persona-aligned behaviors while benefiting from collective knowledge. On the LaMP benchmark, our method improves news categorization F1 by 11.1%, movie tagging F1 by 56.1%, and reduces product rating MAE by 10.4% over prior methods. Our code is available at https://anonymous.4open.science/r/PersonaAgentwGraphRAG-DE6F

PersonaAgent with GraphRAG: Community-Aware Knowledge Graphs for Personalized LLM

TL;DR

The paper tackles the challenge of personalized AI agents by integrating dynamic user personas with community-aware knowledge via GraphRAG. It introduces a Knowledge Graph that encodes individual interactions and community patterns, and a GraphRAG retrieval mechanism that gathers both user-specific and global context to form personalized prompts for LLMs. Empirical results on the LaMP benchmark show consistent improvements across news categorization, movie tagging, and product rating, including notable gains for smaller models. The work demonstrates the value of combining structured memory with graph-based retrieval to achieve grounded, explainable, and adaptable personalization in LLM-driven agents, with future plans toward multi-agent collaboration and IRL-based preference inference.

Abstract

We propose a novel framework for persona-based language model system, motivated by the need for personalized AI agents that adapt to individual user preferences. In our approach, the agent embodies the user's "persona" (e.g. user profile or taste) and is powered by a large language model (LLM). To enable the agent to leverage rich contextual information, we introduce a Knowledge-Graph-enhanced Retrieval-Augmented Generation (Graph RAG) mechanism that constructs an LLM-derived graph index of relevant documents and summarizes communities of related information. Our framework generates personalized prompts by combining: (1) a summary of the user's historical behaviors and preferences extracted from the knowledge graph, and (2) relevant global interaction patterns identified through graph-based community detection. This dynamic prompt engineering approach allows the agent to maintain consistent persona-aligned behaviors while benefiting from collective knowledge. On the LaMP benchmark, our method improves news categorization F1 by 11.1%, movie tagging F1 by 56.1%, and reduces product rating MAE by 10.4% over prior methods. Our code is available at https://anonymous.4open.science/r/PersonaAgentwGraphRAG-DE6F

Paper Structure

This paper contains 14 sections, 3 equations, 3 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Persona-Based LLM Agents
  4. Memory and Knowledge Integration in LLM Systems
  5. Retrieval-Augmented Generation and Knowledge Graphs
  6. Methodology
  7. Knowledge Graph Construction
  8. GraphRAG Retrieval Mechanism
  9. Personalized Prompt Generation
  10. Results
  11. Data Description
  12. Metircs Comparison
  13. Case Study
  14. Conclusion

Figures (3)

  • Figure 1: Overview of the PersonaAgent with GraphRAG framework.
  • Figure 2: LLMs Comparison on LaMP-2N
  • Figure 3: Case study of PersonaAgent with GraphRAG for personalized classification