A Graph Perspective to Probe Structural Patterns of Knowledge in Large Language Models

TL;DR

This paper explores the structural patterns of knowledge in large language models from a graph perspective, revealing insights like knowledge homophily and developing graph-based models for knowledge estimation and improvement.

Contribution

It introduces a novel graph-based analysis of LLM knowledge structures and develops models leveraging local graph properties for knowledge estimation and enhancement.

Findings

Knowledge homophily among topologically close entities

Graph models effectively estimate entity knowledge levels

Fine-tuning with selected triplets improves LLM performance

Abstract

Large language models have been extensively studied as neural knowledge bases for their knowledge access, editability, reasoning, and explainability. However, few works focus on the structural patterns of their knowledge. Motivated by this gap, we investigate these structural patterns from a graph perspective. We quantify the knowledge of LLMs at both the triplet and entity levels, and analyze how it relates to graph structural properties such as node degree. Furthermore, we uncover the knowledge homophily, where topologically close entities exhibit similar levels of knowledgeability, which further motivates us to develop graph machine learning models to estimate entity knowledge based on its local neighbors. This model further enables valuable knowledge checking by selecting triplets less known to LLMs. Empirical results show that using selected triplets for fine-tuning leads to superior performance.