semantic_duplicates 1.0.x-dev

First development snapshot of Semantic Duplicates.

The module finds duplicate and competing content on a Drupal site:
pages copied from other pages, pages that reuse sections of each other,
and separate pages that cover the same topic and compete in search.

What is included

• Lexical detection (shared wording) that works without any AI provider.
• Optional semantic detection using embeddings through the
  AI module.
  It works with a local, self-hosted provider such as Ollama, so content
  does not have to leave your server.
• An audit report of duplicate and competing pages, filtered by
  what each user is allowed to view.
• An advisory warning on the node edit form. It never blocks saving.
• Drush commands for backfilling and inspecting results.
• A demo submodule (sd_demo) with example pages for each
  type of finding.

Requirements

Drupal 11.3 or later, PHP 8.3 or later, and the AI module.

Status

This is a development snapshot, not for production use yet.
Known limitations before the first alpha:

• The edit form warning checks the last saved version of a page,
  so new pages are not checked yet.
• If the AI provider fails, queued items are not retried.
• On multilingual sites, a page can be paired with its own translation.

Feedback and bug reports are very welcome in the issue queue.