GraphRAG Architecture Overview

GraphRAG transforms document collections into structured knowledge graphs, supporting both entity-specific and thematic analysis queries.

Core System Overview

GraphRAG operates through two fundamental processes:

Phase	Purpose	Output
Indexing Pipeline	Analyzes documents to construct structured knowledge	Knowledge Graph + Community Reports
Query Engine	Uses knowledge graph for contextual responses	Contextual Answers

flowchart LR
    A["Document Collection"] --> B["Indexing Pipeline"]
    B --> C["Knowledge Graph<br/>+ Community Reports"]
    C --> D["Query Engine"]
    D --> E["Contextual Responses"]

    classDef inputStyle fill:#f8f9fa,stroke:#6c757d,stroke-width:2px,color:#212529
    classDef processStyle fill:#e3f2fd,stroke:#1976d2,stroke-width:2px,color:#0d47a1
    classDef dataStyle fill:#fff3e0,stroke:#f57c00,stroke-width:2px,color:#e65100
    classDef outputStyle fill:#e8f5e8,stroke:#388e3c,stroke-width:2px,color:#1b5e20

    class A inputStyle
    class B,D processStyle
    class C dataStyle
    class E outputStyle

Indexing Pipeline Architecture

The indexing process turns raw documents into organized knowledge through a step-by-step process:

flowchart TD
    subgraph input ["Input Processing"]
        A["Raw Documents"]
        B["Document Splitting"]
        A --> B
    end

    subgraph extraction ["Knowledge Extraction"]
        C["Entity Recognition"]
        D["Relationship Mining"]
        E["Graph Construction"]
        C --> E
        D --> E
    end

    subgraph organization ["Knowledge Organization"]
        F["Community Detection"]
        G["Summary Generation"]
        F --> G
    end

    subgraph output ["Query Artifacts"]
        H["Searchable Knowledge Base"]
    end

    B --> C
    B --> D
    E --> F
    G --> H

    classDef inputClass fill:#f8f9fa,stroke:#6c757d,stroke-width:2px
    classDef processClass fill:#e3f2fd,stroke:#1976d2,stroke-width:2px
    classDef organizeClass fill:#fff3e0,stroke:#f57c00,stroke-width:2px
    classDef outputClass fill:#e8f5e8,stroke:#388e3c,stroke-width:2px

    class A,B inputClass
    class C,D,E processClass
    class F,G organizeClass
    class H outputClass

Practical Example

Input Document: "Ratan Tata served as Chairman of Tata Group from 1991 to 2012, transforming it into a global business group with acquisitions like Jaguar Land Rover."

GraphRAG Knowledge Extraction

Extract Type	Results
Entities	Ratan Tata • Tata Group • Jaguar Land Rover • 1991 • 2012
Relationships	Ratan Tata → served_as_chairman → Tata Group Tata Group → acquired → Jaguar Land Rover
Communities	Business Leadership • Automotive Industry

Query Engine Architecture

GraphRAG uses two different search methods to handle different types of questions:

Local Search (Entity-Focused Queries)

Aspect	Details
Best For	Specific factual questions about entities and relationships
Examples	"What companies did Ratan Tata lead?" • "When did Tata acquire Jaguar?"
How it Works	Finds entities → Follows connections → Builds context
Characteristics	High precision with specific responses

Global Search (Thematic Analysis)

Aspect	Details
Best For	Big-picture questions requiring complete insights
Examples	"Key business transformation strategies?" • "Leadership patterns in business groups?"
How it Works	Community analysis → Pre-built summaries → Insight combining
Characteristics	Complete coverage with synthesized insights

flowchart TD
    A["User Query"] --> B{"Query Classification"}

    subgraph local ["Local Search Pipeline"]
        C["Entity Resolution"]
        D["Graph Traversal"]
        E["Context Assembly"]
        C --> D --> E
    end

    subgraph global ["Global Search Pipeline"]
        F["Community Matching"]
        G["Report Synthesis"]
        H["Insight Aggregation"]
        F --> G --> H
    end

    I["Response Generation"]

    B -->|"Entity-Specific"| local
    B -->|"Thematic"| global
    E --> I
    H --> I

    classDef queryStyle fill:#f8f9fa,stroke:#6c757d,stroke-width:2px,color:#212529
    classDef decisionStyle fill:#fff3e0,stroke:#f57c00,stroke-width:3px,color:#e65100
    classDef localStyle fill:#e3f2fd,stroke:#1976d2,stroke-width:2px,color:#0d47a1
    classDef globalStyle fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px,color:#4a148c
    classDef outputStyle fill:#e8f5e8,stroke:#388e3c,stroke-width:2px,color:#1b5e20

    class A queryStyle
    class B decisionStyle
    class C,D,E localStyle
    class F,G,H globalStyle
    class I outputStyle

Comparison with Traditional Search

Aspect	Traditional Search	GraphRAG
Understanding	Keyword matching only	Meaningful connections & relationships
Analysis Depth	Single-level results	Detailed (Local) + Strategic (Global) insights
Source Tracking	Basic page references	Complete traceability to original documents
Context Awareness	Isolated results	Connected knowledge with relationships

Knowledge Architecture Components

GraphRAG organizes information into four interconnected layers:

Component	Description	Example
Entities	People, organizations, and concepts	`Ratan Tata` • `Tata Group` • `Jaguar Land Rover`
Relationships	How different entities connect to each other	`Ratan Tata served as Chairman of Tata Group`
Communities	Groups of related entities by topic	`Business Leadership` • `Automotive Industry`
Text Units	Original text pieces with entity links	`"Ratan Tata served as Chairman of Tata Group from 1991..."`

Query Strategy Selection Guide

Choose the right search method for your question type:

Query Example	Search Strategy	Why This Choice
`"What is Ratan Tata's background?"`	Local Search	Entity-specific biographical information
`"Which companies did Tata Group acquire?"`	Local Search	Specific relationship and timeline queries
`"What are the main business transformation patterns?"`	Global Search	Theme analysis across multiple entities
`"Analyze the strategic evolution of Indian business groups"`	Global Search	Complete pattern recognition and insights

Implementation Resources

Resource	Description	Best For
Indexing Pipeline Guide	Complete indexing process documentation	Understanding the build process
Query System Guide	Local vs Global search explained	Learning when to use each query type
System Customization Guide	Component configuration and extensions	Adapting to your needs

Key Value Proposition

GraphRAG transforms how you work with documents by building intelligent knowledge structures that understand context and relationships. This enables both precise factual questions and strategic analytical insights - going far beyond traditional search.

Indexing Pipeline
Technical implementation details and configuration options for building knowledge graphs.

Documentation Index
Return to documentation overview