ES Semantic search demo

This project consists of an Elasticsearch service and a Node.js Express API. The Express API uses an ONNX embeddings model for processing.

Prerequisites

• Docker
• Docker Compose
• Node.js
• Python 3.11

Setup

1. Download the ONNX Embeddings Model

First, run the download_model.py script to download the ONNX embeddings model. This step is necessary for the Express API to function correctly.

python download_model.py

2. Build and Run the Docker Containers

Use Docker Compose to build and run the containers for Elasticsearch and the Express API.

docker-compose up --build

3. Access the Services

• Elasticsearch will be accessible at http://localhost:9200
• The Express API will be accessible at http://localhost:3000

Express API

The Express API provides endpoints to interact with the ONNX embeddings model and Elasticsearch.

Endpoints

• POST /enter-data: Indexes data into Elasticsearch.
• POST /vector-search: Performs a vector similarity search using the ONNX model.
• POST /text-search: Performs a text-based fuzzy search on ProductName and Description.