MongoDB Vector Search Example

This project is a proof-of-concept of using MongoDB's vector search feature, providing sample contents to seed into the database, and a simple API to search them.

Requirements

• MongoDB 7.0 (Right now can be used only on MongoDB Atlas)
• NodeJS

Installation and running

• npm install to installing the dependencies
• Then, copy .env.example to .env and editing the configuration
• Run node seed.js to insert sample data into the database
• Run node server.js to start server
• Try searching with http://localhost:3000?q=<search_term>

Configuration

MONGODB_CONNECTION=mongodb+srv://<username>:<password>@<host>/?retryWrites=true&w=majority
MONGODB_DATABASE=demo
OPENAI_SECRET=secret

• For MongoDB-related config, you can create a new database deployment, and click Connect button for information.
• For OPENAI_SECRET, can be get at API Keys page.

Learn more

• https://www.mongodb.com/docs/atlas/atlas-vector-search/vector-search-overview/#perform-semantic-search-with-atlas-vector-search

Thanks

• Sample news content from Thai PBS