Knowledge Graph Generator

A Colab-based Streamlit application that extracts graph data (entities and relationships) from text input using LangChain and OpenAI's GPT-4o models, then visualizes it interactively as a dynamic knowledge graph.

---

🚀 Features

• 📂 Input support for both .txt uploads and direct text entry
• 🧠 Automatic entity and relationship extraction using GPT-4o
• 🕸️ Interactive knowledge graph with physics-based layout
• 🔄 Real-time drag, zoom, and hover interactivity with PyVis
• 🌐 Fully deployable via Streamlit within Colab using %%writefile

---

⚙️ Installation

✅ Prerequisites

• Python 3.8+
• OpenAI API key

📦 Dependencies

The following Python packages are required:

• langchain >= 0.1.0
• langchain-experimental >= 0.0.45
• langchain-openai >= 0.1.0
• python-dotenv >= 1.0.0
• pyvis >= 0.3.2
• streamlit >= 1.32.0

Install all dependencies using:

pip install -r requirements.txt

In Colab, you can also run:

!pip install -r requirements.txt

---

🔐 API Key Setup

Create a .env file in the notebook with:

%%writefile .env
OPENAI_API_KEY=your_openai_api_key_here

Or set it dynamically in the notebook:

import os
os.environ["OPENAI_API_KEY"] = "your_openai_api_key_here"

---

▶️ Running the App in Colab

Since this app uses %%writefile to generate scripts:

1. Run each notebook cell to write required files like app.py
2. Launch the Streamlit app using:

!streamlit run app.py

Colab will give you a public URL to access the interface.

---

🧠 How to Use

1. Select input method in the sidebar:

   • 📄 Upload a .txt file, or
   • 📝 Enter text directly

2. Click "Generate Knowledge Graph"

3. Interact with the graph:

   • Drag nodes
   • Zoom in/out with mouse wheel
   • Hover to see details
   • Apply graph filters and layouts

---

🧪 How It Works

1. Your input text is sent through a LangChain GraphTransformer
2. GPT-4o identifies (entity) - [relationship] → (entity) triplets
3. A NetworkX graph is built from those triplets
4. PyVis visualizes the graph inside Streamlit

Example triplets:

(Isaac Newton) - [discovered] → (Laws of Motion)  
(Laws of Motion) - [published in] → (1687)

---

📁 Project Structure

All scripts are generated via Colab using %%writefile:

├── KnowledgeGraph.ipynb    # Main notebook
├── app.py                  # Streamlit app
├── utils.py                # Helper logic (if needed) (not in the project yet)
├── .env                    # API key
└── requirements.txt        # Dependencies

---

🧑‍💻 Author

© Aditya Saxena, 2025 Licensed under the MIT License. See: MIT License

---

📚 Resources

• LangChain Docs
• OpenAI API Docs
• Streamlit
• PyVis Docs