Skip to content

noworneverev/graphrag-visualizer

Repository files navigation

GraphRAG Visualizer

πŸ‘‰ GraphRAG Visualizer
πŸ‘‰ GraphRAG Visualizer Demo

demo

Overview

GraphRAG Visualizer is an application designed to visualize Microsoft GraphRAG artifacts. By uploading parquet files generated from the GraphRAG indexing pipeline, users can easily view and analyze data without needing additional software or scripts.

Important Note

If you are using GraphRAG 0.3.x or below, please use the legacy version of GraphRAG Visualizer available at:
πŸ‘‰ GraphRAG Visualizer Legacy

Features

  • Graph Visualization: View the graph in 2D or 3D in the "Graph Visualization" tab.
  • Data Tables: Display data from the parquet files in the "Data Tables" tab.
  • Search Functionality: Fully supports search, allowing users to focus on specific nodes or relationships.
  • Local Processing: All artifacts are processed locally on your machine, ensuring data security and privacy.

Using the Search Functionality

Once the graphrag-api server is up and running, you can perform searches directly through the GraphRAG Visualizer. Simply go to the GraphRAG Visualizer and use the search interface to query the API server. This allows you to easily search and explore data that is hosted on your local server.

search

Graph Data Model

The logic for creating relationships for text units, documents, communities, and covariates is derived from the GraphRAG import Neo4j Cypher notebook.

Nodes

Node Type
Document RAW_DOCUMENT
Text Unit CHUNK
Community COMMUNITY
Finding FINDING
Covariate COVARIATE
Entity Varies

Relationships

Source Node Relationship Target Node
Entity RELATED Entity
Text Unit PART_OF Document
Text Unit HAS_ENTITY Entity
Text Unit HAS_COVARIATE Covariate
Community HAS_FINDING Finding
Entity IN_COMMUNITY Community

Developer Instructions

Setting Up the Project

  1. Clone the repository to your local machine:

    git clone https://github.com/noworneverev/graphrag-visualizer.git
    cd graphrag-visualizer
  2. Install the necessary dependencies:

    npm install
  3. Run the development server:

    npm start
  4. Open the app in your browser:

    http://localhost:3000
    

Loading Parquet Files

To load .parquet files automatically when the application starts, place your Parquet files in the public/artifacts directory. These files will be loaded into the application for visualization and data table display. The files can be organized as follows:

  • public/artifacts/create_final_entities.parquet
  • public/artifacts/create_final_relationships.parquet
  • public/artifacts/create_final_documents.parquet
  • public/artifacts/create_final_text_units.parquet
  • public/artifacts/create_final_communities.parquet
  • public/artifacts/create_final_community_reports.parquet
  • public/artifacts/create_final_covariates.parquet

If the files are placed in the public/artifacts folder, the app will automatically load and display them on startup.