GraphRAG

GraphRAG experimentation project for building and querying graph-structured knowledge from scraped content and extracted entities/relations.

Repository Structure

• graphrag/
  • Core package modules for config, extraction, prompts, storage, querying, and visualization
• outputs/
  • Centralized generated artifacts (scraped data, schema suggestions, graph files, persisted storage)
• pipeline.py
  • End-to-end pipeline orchestration
• GraphEngine.py
  • Graph construction and processing logic
• scrape.py
  • Data collection and scraping flow
• schema.json
  • Active schema source of truth used by runtime config
• graph_template.html
  • Visualization template

Main Artifacts

• outputs/scraped_results.csv
• outputs/schema/schema_candidates.json
• outputs/schema/schema_suggested.json
• outputs/graph/graph_data.json
• outputs/graph/ai_graph.html
• outputs/storage/community_summaries.json
• outputs/storage/graph_structure.json

Quick Start

1. Create and activate a Python virtual environment.
2. Install required dependencies used by this project.
3. Run scraping (optional):

python scrape.py

4. Run pipeline:

python pipeline.py

5. Open generated graph HTML outputs to inspect results.

Notes

• Keep secrets/API keys in local environment files only.
• This repository ignores local virtual environments and .env files.