Theka v1.1.5 — Advanced multi-field search

Smarter search

• Unified search across all metadata: title, protocol number, external code, category, type, tags, filename, file type, status, notes and custom data are all considered in a single search
• AI-generated summary, description, tags and extracted data are also indexed: find content even when you don't remember the exact words of the document
• Document contents (PDF, Word, Excel) are searched in parallel with metadata, without hurting performance even on files with hundreds of pages

Search in linked entities

• Find a document starting from the linked supplier/customer data: name, external code, VAT number, tax code, address or any other custom entity field
• Real-time updates: update a supplier's VAT number? The next search finds the linked documents updated instantly, no waiting time
• Same logic for projects: search by project name, description or external code

Grouping and match context

• Results are grouped by document: if the match was found in an older version while a newer one exists, you get a clear notice
• Every result shows where the match was found: in the metadata, in the document content, by meaning, in the linked entity or in the project
• When the match is on an entity, the entity name and the specific field that caused the match are shown (e.g. "Acme Inc. — VAT number")

New filters

• "Also search archived" option: archived versions are excluded by default but can be included when needed
• "Also search deleted" option (admin only): useful for data recovery and audits

• Result ranking has been redesigned with weighted scoring: matches on title and protocol number weigh more than matches on notes, and the current version takes precedence over older versions
• Changing category, type, title, protocol number or external code of a document now automatically updates the search index for all its versions
• Improved performance on searches that include entity custom data (VAT number, codes, etc.) thanks to a new dedicated index
• Search results always include information about the current version and the one where the match was found, to avoid consulting outdated data

• Search is no longer limited to the current version of a document: if the searched data exists in a previous version, it is found and clearly flagged
• The content of a very long document no longer slows down the general system search

Update focused on search improvements. Expected outcomes: more relevant results, less noise, a better experience with the AI assistant. No changes to existing data.

After the update a few minutes may be needed for search indexes to update on existing documents.