Semantic Search - AI Vectored Search

The Drupal Semantic Search module makes it possible to enhance your Search engine using AI tools.

The aim of this module is to integrate with various AI libraries to leverage AI-powered search functionnalities. While the OpenAI API module already offers semantic search faculties thanks to its embedding submodule, it is tightly ties (as its name suggests) to the OpenAI ecosystem. This is rather limiting and rises GDPR and privacy issues, since all data is being sent to a private company. We solve this limitation by opening the vectorization and embedding storage options to other vendors.

For a full description of the module, visit the
project page.

Submit bug reports and feature suggestions, or track changes in the
issue queue.

Table of contents

• Requirements
• Installation
• Webserver Streaming Support
• Planned Functionality
• Maintainers

Requirements

This module is tested on Drupal 10.x.

Installation

Enable the Semantic Search module and one or more submodules that meet your need.

Vectorization

We currently offer two options for generating embeddings.

OpenAI

You need to provide an OpenAI key.

Custom API

You need to provide the URL and Authorization Token to your custom model.

Embedding Storage

3 options are now available.

Local Storage

Objects are stored locally in a JSON file. No requirements here.

Postgres Server

You need access to a Postgres server (version 13+) with the Vector extension enabled.

Pinecone

You need to create an index, with a vector size equal to your vectorization requirements (e.g. OpenAI is 1536).

Planned functionality

This module is under active development, and all suggestions / contributions are welcomed. In the meantime, we plan the following developments :

• Search API Integration
• More connectors to models
• More connectors to vector storage options
• Content moderation
• Views integration
• and more

Maintainers

• Guillaume Cambon - oeklesund
• Alain Gabelle - AlGab