README

`sndsabin/module-embedding`

• Main Functionality
• Installation
• Configuration
• Generate Product Embedding
• Tested On

Main Functionality

This module generates product embeddings for Magento 2 catalogs using OpenAI-compatible models.

⚠️ This is not a standalone module. It is intended to be used together with a Semantic Search module.

Installation

• Install the module via composer:

composer require sndsabin/module-embedding

• Enable the module:

bin/magento module:enable SNDSABIN_Embedding
bin/magento setup:upgrade

Configuration

Navigate to Stores → Configuration → Embedding → Settings

General

• API Key
  • Required for most providers
  • If your provider does not require a key, enter any placeholder value

Advanced

• Base URI of Model

  • Example: https://api.openai.com/v1

• Model

  • Example: text-embedding-3-small

• Embedding Dimension

  • Example: 1536 (for text-embedding-3-small)

• Index Name

  • Example: product_embedding_v1
  • When changing model or dimensions, update index version
    (e.g. product_embedding_v1 → product_embedding_v2)

• Attributes for Embedding

  • Product attributes used to generate embeddings
  • Note: Name, SKU, Short Description, and Description are always included

⚠️ After configuration, Generate product embeddings. This process may take some time depending on catalog size.

Generate Product Embedding

1. Queue embedding generation

bin/magento sndsabin:catalog_product:embedding:generate

This command queues embedding jobs for all products.

⚠️ This command is usually run once during initial setup.

Embeddings for new or updated products are generated automatically.

2. Run the consumer

bin/magento queue:consumers:start catalogProductEmbeddingConsumer --batch-size=100

This consumer processes queued messages and generates embeddings. Use cronjob or supervisor to start the consumer.

Tested On:

• Magento 2.4.8
• PHP 8.4
• RabbitMQ 4.1
• OpenSearch 3.3