In content-heavy sites (like those our clients tend to have) there is typically a curated list of different vocabularies – and terms within those vocabularies – that have been painstakingly reviewed and thought through to best support end-users in finding content through filterable listing pages, search, related content and other means.
When an individual author tries to keep all the possible categorisations in their head, or reviews each one manually when creating content, it can become an inefficient process and one that is prone to error.
This particular task is a job that AI can efficiently take care of as there is a finite number of terms and a clear chunk of content ready to be compared. It just needs to understand the language, closeness of terms, and context – something at which large language models (LLMs) like ChatGPT excel.
We have built a tool on top of OpenAI’s API (ie, the same API that ChatGPT uses) that does just that. And we’ve made it readily available for all our Drupal clients. You can see it in action here.
Efficient tagging with AI precision
The AI auto-tagging system is designed to distinguish between varying levels of relevance. It identifies the ‘highest relevant terms’ and ‘medium relevant terms’ for each piece of content.
- Highest relevant terms
These are the terms that the AI system identifies as most closely associated with the content. They are selected by default, allowing you to apply these tags with a single click. This significantly speeds up the tagging process. - Medium relevant terms
These terms, while relevant, are not as closely aligned with the content as the highest relevant terms. They are suggested for additional consideration, giving site authors and editors the flexibility to choose how extensively to tag the content.
Increasing confidence and automation over time
As you use the system more and become confident in its accuracy, there is an option to skip the review and approval process altogether and let the AI automatically apply, for example, the high relevance tags.
Interested in finding out more? Drop us a line at hello@designbysoapbox.com.