Skip to content

README.md

Latest commit

 

History

History
69 lines (41 loc) · 2.17 KB

README.md

File metadata and controls

69 lines (41 loc) · 2.17 KB
Raw

🧠 RAG & GraphRAG Project

This repository contains implementation and benchmark of retrieval-augmented generation (RAG), graph-based document analysis (GraphRAG).

Project Structure

RAG_OpenAI/

This folder contains a complete implementation of a classic RAG system using OpenAI embeddings, ChromaDB, and LangChain.

  • Loads .txt, .pdf (and optionally .md) documents
  • Stores them in a vector database
  • Enables conversational retrieval using GPT-4o

A detailed README is included in the folder to guide you step-by-step on usage, configuration, and extensions.

Link for a direct access detailed README for the RAG module.

GraphRAG_Microsoft/ (from Microsoft)

This folder contains :

  • The full instructions to implement your graphRAG
  • A custom code to help visualize the Graph.

A detailed README is included in the folder to guide you step-by-step on usage, configuration, and extensions.

Link for a direct access detailed README for the GraphRAH module.

Results excel/

This directory contains the evaluation results of the system using:

  • Human annotations
  • LLM-based scoring
    Stored in .xlsx format.

Data location/

This folder includes .csv files listing the URLs of public articles used as data sources for the report, covering:

  • The 2016 election of Donald Trump
  • The Russia–Ukraine conflict

The data was retrieved using TheGuardian's official API.

Summary

Folder Description
RAG_OpenAI/ Full RAG pipeline with OpenAI + ChromaDB + LangChain
GraphRAG/ Microsoft GraphRAG integration + Gephi visualization support
Results excel/ XLSX files of human and LLM-based evaluation results
Data location/ CSV lists of article URLs used as data sources

Detailed explanation

Just read the Project_report.pdf to see the process and the results.

Link for a direct access Project_report.pdf.