Knowledge Graph Builder

Build and visualize knowledge graphs from PDF documents using NLP. No external APIs required.

Quick Start

./run.sh          # macOS/Linux
# or
run.bat           # Windows

The system will:

1. Extract text from PDFs in src/docs/
2. Build a knowledge graph using NLP (SpaCy)
3. Start server at http://localhost:8000

Features

• Automatic Graph Building - Extracts entities and relationships from PDFs
• Interactive Visualization - D3.js powered graph with zoom, pan, and drag
• Dual Storage - JSON (default) or Neo4j for large-scale graphs
• Auto-Rebuild - Fresh graph on every run

Installation

# Create virtual environment
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Download NLP model
python -m spacy download en_core_web_sm

# Run
python src/main.py

Usage

Add Documents

Place PDFs in src/docs/ and run the system

View Graph

http://localhost:8000/visualize

API Endpoints

• /api/graph - Graph data
• /api/stats - Statistics
• /docs - API documentation

Neo4j (Optional)

# Create .env file
cp .env.example .env

# Edit .env
USE_NEO4J=true
NEO4J_URI=bolt://localhost:7687
NEO4J_USERNAME=neo4j
NEO4J_PASSWORD=your-password

Tech Stack

• NLP: SpaCy for entity extraction
• Graph: NetworkX
• Storage: JSON / Neo4j
• API: FastAPI
• Frontend: D3.js

Design

Follows SOLID, KISS, and YAGNI principles with clean separation of concerns.