Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Explore greater use of automatic translations in Weblate #66

Open
hturner opened this issue Sep 23, 2024 · 0 comments
Open

Explore greater use of automatic translations in Weblate #66

hturner opened this issue Sep 23, 2024 · 0 comments
Labels
SIP 2024 Items reserved for R Dev Day @ SIP 2024

Comments

@hturner
Copy link
Member

hturner commented Sep 23, 2024

So far we have only been using machine translation to suggest translations when translating one string at a time.

It seems that its is possible to bulk translate a whole component:

  • Navigate to the translations for a particular component in a particular language, e.g. https://translate.rx.studio/projects/r-project/base-c/ar/
  • Select 'Tools > Automatic translation'
  • In the dialog,
    • under 'Automatic translation mode', select 'Add as "needing edit"'
    • under 'Search filter', select 'Untranslated strings'
    • under 'Source of automated translations', select 'Machine translation'
    • under 'Machine translation engines', select 'Microsoft Translator'

Other options are available and could be explored. By selecting 'Add as "needing edit"', the autotranslated strings would be treated as "fuzzy" translations, i.e. they would be added to the component source files when a patch is made, but they would NOT be used. My rationale is that a translator should review the translations, edit as necessary and uncheck "needing edit". Possibly teams using a dedicated review process (where strings require approval) can add select 'Add as translation', but this would need to be carefully checked, e.g. on a component with few untranslated strings.

Microsoft Translator is preferred for bulk translation as our free tier covers 2 million characters/month. DeepL only allows 500k chars/month, so it may be better to save this for translating one string at a time, but it could be tried if Microsoft Translator is much worse

@hturner hturner added the SIP 2024 Items reserved for R Dev Day @ SIP 2024 label Sep 27, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
SIP 2024 Items reserved for R Dev Day @ SIP 2024
Projects
None yet
Development

No branches or pull requests

1 participant