Full-text search
Search & navigation ›Full-text search
April 7, 2026Updated: April 7, 20265 min read23 views
How to find documents by searching words in content
Full-text search
Theka indexes the text content of all documents: title, notes, extracted text, custom fields, tags and AI results.
How it works
- Understands Italian and English (stemming: "contracts" also finds "contract", "contractual")
- Ignores articles, prepositions and stop words
- Ranks results by relevance (how often and where the term appears)
- Prioritizes: current version, final/signed status, active project
Search syntax
| You type | What it searches | Example |
|---|---|---|
organic vegetables | Both words (implicit AND) | Documents with "organic" AND "vegetables" |
"monthly supply" | Exact phrase in quotes | Only the exact expression |
supply OR consulting | Either word | Documents with "supply" OR "consulting" |
contract -draft | First word BUT NOT the second | Contracts that are not drafts |
"delay penalty" supplier | Exact phrase + other word | Delay penalties in supplier documents |
Where it searches
Full-text search looks in:
- Document title
- Version notes
- Text extracted from file (PDF, Word, Excel)
- Tags (manual and AI)
- Custom field values (custom_data)
- AI summaries and descriptions (if available)
Combinable filters
You can combine text search with structured filters:
- Category and Document type
- Status (draft, review, final, signed)
- Direction (active, passive, internal)
- Entity (supplier, client, employee)
- Project
- Tags
- Date range
⚡ Performance: PostgreSQL GIN indexes, results in milliseconds even with thousands of documents.
💡 Tip: Full-text search always works, no AI configuration needed. If you're looking for something but don't remember the exact words, try semantic search.