Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TERAG: Token-Efficient Graph-Based Retrieval-Augmented Generation

Qiao Xiao, Hong Ting Tsang, Jiaxin Bai

TL;DR

This work proposes TERAG, a simple yet effective framework designed to build informative graphs at a significantly lower cost, and incorporates Personalized PageRank (PPR) during the retrieval phase, and achieves at least 80% of the accuracy of widely used graph-based RAG methods while consuming only 3%-11% of the output tokens.

Abstract

Graph-based Retrieval-augmented generation (RAG) has become a widely studied approach for improving the reasoning, accuracy, and factuality of Large Language Models (LLMs). However, many existing graph-based RAG systems overlook the high cost associated with LLM token usage during graph construction, hindering large-scale adoption. To address this, we propose TERAG, a simple yet effective framework designed to build informative graphs at a significantly lower cost. Inspired by HippoRAG, we incorporate Personalized PageRank (PPR) during the retrieval phase, and we achieve at least 80% of the accuracy of widely used graph-based RAG methods while consuming only 3%-11% of the output tokens. With its low token footprint and efficient construction pipeline, TERAG is well-suited for large-scale and cost-sensitive deployment scenarios.

TERAG: Token-Efficient Graph-Based Retrieval-Augmented Generation

TL;DR

This work proposes TERAG, a simple yet effective framework designed to build informative graphs at a significantly lower cost, and incorporates Personalized PageRank (PPR) during the retrieval phase, and achieves at least 80% of the accuracy of widely used graph-based RAG methods while consuming only 3%-11% of the output tokens.

Abstract

Graph-based Retrieval-augmented generation (RAG) has become a widely studied approach for improving the reasoning, accuracy, and factuality of Large Language Models (LLMs). However, many existing graph-based RAG systems overlook the high cost associated with LLM token usage during graph construction, hindering large-scale adoption. To address this, we propose TERAG, a simple yet effective framework designed to build informative graphs at a significantly lower cost. Inspired by HippoRAG, we incorporate Personalized PageRank (PPR) during the retrieval phase, and we achieve at least 80% of the accuracy of widely used graph-based RAG methods while consuming only 3%-11% of the output tokens. With its low token footprint and efficient construction pipeline, TERAG is well-suited for large-scale and cost-sensitive deployment scenarios.

Paper Structure

This paper contains 35 sections, 9 equations, 3 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Naive RAG
  4. Graph RAG
  5. The Challenge in Graph RAG
  6. Problem Definition
  7. Token Consumption
  8. Graph Definition
  9. Graph Type
  10. Node Types
  11. Edge types
  12. Objective
  13. Pipeline
  14. Named Entity and Document-Level Concept Extraction
  15. Named Entities
  16. ...and 20 more sections

Figures (3)

  • Figure 1: Overall pipeline of the proposed TERAG framework. The process consists of lightweight concept extraction with LLMs, followed by efficient non-LLM clustering and graph construction.
  • Figure 2: Graph structure of TERAG. The figure illustrates how lightweight concept extraction with LLMs, followed by non-LLM clustering and graph construction, leads to an efficient knowledge graph.
  • Figure 3: Accuracy versus output token consumption on the 2Wiki dataset. The upper-right region indicates higher accuracy with lower token consumption.