NebulaGraphQAChain

Overview

NebulaGraphQAChain is a question-answering framework based on NebulaGraph, LangChain, and DeepSeek (SiliconFlow). It enables querying and retrieving structured knowledge stored in a Nebula Graph database using an AI model.

Environment Configuration

The following environment is used:

• Operating System: Ubuntu 20.04
• Graph Database: Nebula Graph 2025.01.13-nightly
• Programming Language: Python 3.10.16

Installation & Setup

1. Clone the Repository

# Clone the repository to your local machine
git clone https://github.com/EvannZhongg/NebulaGraphQAChain
cd NebulaGraphQAChain

Or you can just download this repository and run it.

Relationship Visualization.py is just a test and does not involve the use of NebulaGraphQAChain. You can delete it.

2. Install Dependencies

Make sure you have Python installed, then install the required dependencies:

pip install -r requirements.txt

3. Configure API Credentials

Edit NebulaGraphQAChain.py and .env to set up your API credentials and model.

API Configuration

Update the API URL and API Key in .env based on your usage:

# SiliconFlow online API call method (modifiable as needed)
CHAT_API_URL=https://api.siliconflow.cn/v1/chat/completions
API_KEY=*****************************  # Replace with your API key

Model Selection

Modify the model you need (default: deepseek-ai/DeepSeek-V3):

response = requests.post(
    CHAT_API_URL,
    headers={"Authorization": f"Bearer {API_KEY}"},
    json={
        "model": "deepseek-ai/DeepSeek-V3",  # Change to the desired model
        "messages": [{"role": "user", "content": prompt}]
    }
)

4. Configure Nebula Graph Connection

Modify the Nebula Graph connection settings in NebulaGraphQAChain.py:

from nebula3.gclient.net import NebulaGraph

graph = NebulaGraph(
    space="YOUR_SPACE",  # Change to the Nebula Graph space you need
    username="root",
    password="nebula",
    address="127.0.0.1",
    port=9669,
    session_pool_size=30,
)

Running the Application

After configuring the necessary settings, run the application:

python NebulaGraphQAChain.py

Usage

1. Open the Web Interface

Once the script is running, open your browser and enter:

http://localhost:5000/

2. Start a Conversation

Enter your questions in the web interface, and the system will generate responses based on the Nebula Graph database and AI model.

3. Monitor Logs

You can view the backend log on the frontend page.

Backend Log View:

nGQL Query Execution:

To inspect the generated nGQL queries, check the terminal:

4. Verify Queries in Nebula Graph

Queries can be directly executed in Nebula Graph for validation:

Contribution

If you wish to contribute, feel free to fork the repository and submit a pull request.

License

This project is licensed under the MIT License.

Additionally, this project makes use of LangChain and NebulaGraphQAChain, which is licensed under the MIT License.