Content Matcher

A web application that uses vector-based similarity search to find courses and events that best match a user's provided description. This application uses Next.js, MongoDB Atlas, LangChain, and OpenAI embeddings.

This is a Next.js project bootstrapped with create-next-app.

Setup

Prerequisites

• Node.js (v18 or later)
• MongoDB Atlas account
• OpenAI API key

Environment Variables

Copy the env.sample file to .env.local and fill in your MongoDB Atlas connection string and OpenAI API key:

cp env.sample .env.local

Then edit .env.local with your actual credentials:

MONGODB_URI=mongodb+srv://<username>:<password>@<cluster>.mongodb.net/<database>?retryWrites=true&w=majority
OPENAI_API_KEY=your_openai_api_key_here
EVENT_VECTOR_SEARCH_INDEX_NAME=event_vector_index
COURSE_VECTOR_SEARCH_INDEX_NAME=course_vector_index

MongoDB Atlas Vector Search Setup

1. Create a MongoDB Atlas cluster if you don't have one already

2. Create a database with the following collections:

   • events and event_embeddings for event data
   • courses for course data (includes embedded vectors)

3. Set up vector search indexes with the following configurations:

   For Events:

   • Index name: event_vector_index (or whatever you specified in EVENT_VECTOR_SEARCH_INDEX_NAME)
   • Collection: event_embeddings
   • Field to index: embedding
   • Dimension: 3072 (for OpenAI's text-embedding-3-large model)
   • Metric: cosine

   For Courses:

   • Index name: course_vector_index (or whatever you specified in COURSE_VECTOR_SEARCH_INDEX_NAME)
   • Collection: courses
   • Field to index: embedding
   • Dimension: 3072 (for OpenAI's text-embedding-3-large model)
   • Metric: cosine

Installation

Install the dependencies:

npm install

Getting Started

Seed the Database

To populate the database with sample content and generate embeddings:

For Events:

npx ts-node src/scripts/seed-events.ts

For Courses:

npx ts-node src/scripts/seed-courses.ts

Local Database (recommended for testing)

To test without touching production, run a local MongoDB with Atlas Vector Search in Docker (docker-compose-local.yml, uses the mongodb/mongodb-atlas-local image).

First, point .env.local at the local database (not your Atlas/production URI):

MONGODB_URI=mongodb://localhost:27018/event-matcher?directConnection=true

Then:

npm run db:up      # start the local database (host port 27018, override with MONGODB_LOCAL_PORT)
npm run db:init    # create collections and vector search indexes (first time only)
npm run seed       # seed local data
npm run db:down    # stop (data persists); `npm run db:reset` also deletes the data

The seed scripts delete existing data, so they refuse to run unless MONGODB_URI points to a local host. Set ALLOW_REMOTE_DB=true to override this deliberately.

Run the Development Server

npm run dev

Open http://localhost:3000 with your browser to see the application.