Zero level support

This project offers an alternative to traditional search engines like Elastic Search, leveraging the power of OpenAI embeddings to enhance the search functionality. It transforms queries into embeddings and searches a PostgreSQL database for the most similar results. This solution features synonym management and special flags to trigger custom behaviors when articles with these flags are identified as similar.

Warning

This project is still in the early stages of development and is not meant to be used in production.

Features

Embeddings-Based Search: Utilizes OpenAI embeddings to understand and match queries with similar content in the database.
Synonym Support: Incorporates a synonym feature to recognize and treat different terms or phrases as equivalent in search queries.
Custom Flags: Allows articles to be tagged with special flags, enabling custom actions or filters when these articles are matched.
Telegram Bot: Includes an example Telegram bot for interactive querying, built on a Django backend with a fully functional admin panel. Web Interface: Provides a simple web-based interface for knowledge base viewing and editing, making content management straightforward.
FastAPI Endpoints: Offers FastAPI endpoints for webhooks, bot status, and similar searches, facilitating integration with other services.
Dockerized Deployment: Designed to run in a Docker Compose environment with PostgreSQL, ensuring easy setup and scalability.

Getting Started

1. Clone the Repository

git clone https://github.com/cyrillsemenov/ZeroLevelSupport.git
cd ZeroLevelSupport

2. Configuration

Copy and fill .env file.

cp .env.sample .env

You will need:

SECRET_KEY — Django secret key
Database config — db name, user, password, host, and port for your database
HOST — this is your webhook host
OPENAI_API_KEY - API key for your OpenAI account
BOT_API_KEY - Telegram bot API key, which you can get from @Botfather
WEB_APP_URL - URL for the shutdown report form

Tip

If you're running this project locally, you'll likely want to set up something like ngrok to obtain your webhook host.

3. Build and Run with Docker Compose

docker-compose up --build

4. Accessing the Admin Panel

Navigate to http://localhost:8000/admin to access the Django admin panel. Use the credentials provided during the setup to log in.

5. Interacting with the Telegram Bot

After setting up the Telegram bot with @Botfather, you can start querying your knowledge base directly from Telegram.

6. Using the Web Interface

Visit http://localhost:8000/questions to view, search, and edit the knowledge base articles through the web interface.

7. API Usage

The FastAPI documentation available at http://localhost:8000/docs details the available endpoints for webhooks, bot status, and similar searches.

8. Contributing

Contributions to enhance this project are welcome! Please feel free to fork the repository, make your changes, and submit a pull request.