This module is intended to use AI to automatically suggest relationships to other content within the same website. As a site builder, you choose whether to auto-accept vs have a review step. Typically this would be automatically tagging/categorising content.

This module integrates with AI core to allow you to use any model.

Screencast showing the module suggesting terms based on the contents.

Features

  • Set up automatic references for any entity reference field (whether it be to taxonomy terms, or other nodes)
  • Automatically first shortens the content if needed in order to fit within token limits of the API
  • Adjust your token limit (eg, if you have been accepted for a higher token limit)
  • Decide whether you wish to review suggestions from the AI before saving the relationships to your content, or whether to auto-apply.
  • Choose different versions of OpenAI models or other models all via the AI Core module.

Setup

  1. Configure the settings page, choosing your OpenAI (or other) version and setting your token limit
  2. Optionally edit the default prompts, or create your own
  3. Edit the content type, go to the 'AI Auto-reference' tab, and decide which fields to automatically generate suggestions for

Post-Installation

  • Configure the AI module to have a Chat provider set up, for example OpenAI
  • Use the settings form to configure your preferences (e.g. review step vs auto-apply).
  • Determine per node bundle which entity reference fields you would like to have the AI suggest results for.

Similar projects

The excellent OpenAI module automatically suggests tags; however, it does not suggest them from the existing taxonomy terms or content entities within a site. For an organisation that has a curated set of categories that they wish to maintain, this module may be more suitable than that particular feature (which is just one of many features OpenAI offers).

Supporting organizations: 
Fully sponsored initial development and ongoing maintenance
Did the majority of development after proof of concept
AttachmentSize
screencast-suggestion-references.gif797.83 KB

Project information

Releases