Review extracted glossary terms
Have Glossa propose glossary terms from your source strings, then approve the ones that belong in your glossary.
An extraction reads your project's source strings, finds the terms that recur, and proposes the ones worth pinning as glossary candidates. Nothing is added on your behalf: a candidate is not enforced until someone approves it.
Who can extract
Owners, Admins and Developers, on Team and above. An extraction uses your AI allowance, and each plan includes a set number of extraction runs per cycle.
Run an extraction
Open the candidates
Open the project's Glossary and select the Review candidates tab.
Extract terms
Select Extract terms. The notice reads "Extraction queued. Terms appear under Review candidates as the run finds them."
An extraction only reads source strings that changed since the last one, so running it again on an unchanged project finds nothing new and costs nothing. Terms already in the glossary, in any state, are never proposed again. When there is nothing new, the notice reads "Nothing to extract. No new candidate terms were found in the source strings."
Review the candidates
Candidates are listed under Waiting for your decision, most confident first. Each shows the term, its kind of word where one was found, and how confident the extraction is that it belongs in a glossary.
- Approve moves the term to the Terms tab, where it is matched and enforced from the next translation: "“Checkout” approved. It is now under Terms."
- Reject drops it: "“Checkout” rejected. It will not be proposed again."
A candidate can carry suggested wording for your target languages. The candidate list does not show it; once the term is approved, its wording appears in the language columns of the Terms tab. Check it there, and to change it, import a file that includes the term.
From the API
A deploy token with the manage ability can start an extraction with POST https://api.transglot.ai/v1/glossary/extract. A token with the read ability can list candidates with GET https://api.transglot.ai/v1/glossary?status=pending. See the API reference.
If it doesn't work
- "Nothing awaiting review": no extraction has run, or you have already decided on everything it found.
- "Candidate extraction unlocks on the Team plan.": extraction is not on your plan. You can still add terms yourself.
- No Extract terms button: your role cannot change the glossary. Ask a Developer, an Admin or an account Owner.