Chatbot Omeka Classic plugin

Releases

1.0.0

Oct 27, 2026

First version for Omeka Classic, the counterpart of the Chatbot 1.5 module for Omeka S.

  • Chat with the collection: a floating button on every public page and the [chatbot] shortcode for simple pages and exhibits, with numbered citations linking to the records, optionally kept to some collections. Answers come from an OpenAI chat model reading the closest passages of the index only.
  • Semantic search: a search page at /semantic-search (optionally in the main navigation) and the [semantic_search] shortcode, one result per page with a score and a highlighted passage, with shareable addresses.
  • The index, in Turso (libSQL with a DISKANN vector index; the same table as the Omeka S module's): public items with their metadata, item type, collection and tags; files with metadata of their own; collections; published simple pages; pages of public exhibits. Optionally only some collections.
  • Kept current: saved records are queued and indexed by one background job (an import starts one job, not one per item); deleted and private records leave the index at once; making an item public again re-indexes its files. Re-index everything, some kinds of records, or the items chosen in batch edit.
  • Admin: a dashboard (chats, searches, answer times, errors, rate limits, a 30-day chart, the latest questions with their answers and sources, the most asked), settings, an Index tab (state, re-index, test search, remove non-public chunks, purge), and a log with level filter and CSV export, kept for a set number of days.
  • Security: the OpenAI key and Turso token are stored encrypted with a key derived from db.ini (or chatbot.encryption_key), or set in config.ini; the system prompt is set in the admin only; chat and search are rate limited per IP address, with forwarded addresses trusted only from local proxies; errors detail goes to the log, not to visitors.
  • Any service with OpenAI's API can be used (the API address setting or chatbot.openai_base_url).
Log in for Support