🧠 Knowledge Graph From Text

A Streamlit web app that generates interactive knowledge graphs from plain text using Ollama AI models. You can upload a .txt file or paste text, and the app will create a graph where concepts are visualized as connected nodes.

---

📸 Preview

🚀 Features

• Two input methods: Upload .txt file or paste text directly.
• Ollama model integration: Select from available local models (e.g., Gemma, Mistral, LLaMA).
• Automatic graph storage: Generated graphs are saved and can be reloaded anytime.
• Interactive visualization: Zoom, drag, and explore relationships between concepts.
• Optimized for speed: Uses hashed filenames to prevent regenerating the same graph.

---

📂 Project Structure

.
├── app.py            # Main Streamlit app 
├── Data              # Folder containing .html files.
├── n_sc.png          # Image file (doc)
├── README.md         # Project documentation and instructions(doc)
├── requirements.txt  # List of Python dependencies for the project(libraries)
├── sc.png            # Another image file (doc)
└── src
    ├── config
    │   └── folder_con.py       # Configuration for folder creation
    ├── graph
    │   ├── generate_kgraph.py  # Script to generate knowledge graphs
    │   └── visulization.py     # Script to visualize knowledge graphs (PyVis)
    ├── model
    │   └── model_info.py       # information about installed Ollama models/context len
    └── utils
        ├── file_op.py          # Utility functions for file operations (read/write)
        └── text_clean.py       # Utility functions for text preprocessing/cleaning

---

⚙️ Installation

A. Non-Docker Installation

1. Clone the repository

git clone https://github.com/ganeshnikhil/Kgraph.git
cd Kgraph

2. Create a virtual environment

python -m venv venv
source venv/bin/activate  # On Linux/Mac
venv\Scripts\activate     # On Windows

3. Install dependencies

pip install -r requirements.txt

4. Install Ollama (if not already installed) Follow instructions from: https://ollama.ai/download

5. Start Ollama server

ollama serve

6. Download a model (example with Gemma)

ollama pull gemma3:4b

7. Run the Streamlit app

streamlit run app.py

---

B. Docker Installation

1. Clone the repository

git clone https://github.com/ganeshnikhil/Kgraph.git
cd Kgraph

2. Build Docker images

docker-compose build

3. Start services (Ollama + Streamlit)

docker-compose up -d

4. Pull Ollama models inside the container

docker exec -it ollama ollama pull <model-name>

5. Access the Streamlit app Open your browser and go to: http://localhost:8501

---

▶️ Usage

• Upload a .txt file or paste your text in the sidebar.
• Select an Ollama model to use for generating the graph.
• Click "Generate Knowledge Graph".
• View, zoom, and explore the interactive visualization.
• Load previously generated graphs from the sidebar.

---

📌 Notes

• All generated graphs are saved in the Data/ directory as .html files.
• Large input text may take more time to process depending on the model.
• If no model is found, the app will fall back to Gemma.

---

📜 License

This project is licensed under the MIT License — see the LICENSE file for details.