Page MenuHomePhabricator

Check Apertium configuration for Serbo-croatian
Closed, ResolvedPublic

Description

The configuration file for Apertium indicates that it supports the English(en) to Serbo-croatian (sh) language pair. However, that language does not appear as supported in the Apertium website, and produces lot of errors when used:

Screenshot 2020-10-02 at 11.46.33.png (401×1 px, 154 KB)

The purpose of this ticket is to check the configuration to confirm whether the corresponding en-sh package exist, and update the configuration accordingly.

Event Timeline

  • eng-hbs (English -> Serbo-croation) pair is installed on our server.
  • hbs is correctly mapped to 'sh' in mapping file. (cxserver/lib/mt/Apertium.languagenames.json)
  • Translation quality is very low for this pair in Apertium. We can disable this pair for Apertium, if required.

Thanks for checking and reporting on the current situation, @KartikMistry.
I think that an immediate step would be making Google the default service for Serbo-croatian.

In that way, Apertium can be still accessed to those really interested (e.g., for checking if improves in the future).

Change 632646 had a related patch set uploaded (by KartikMistry; owner: KartikMistry):
[mediawiki/services/cxserver@master] Make Google MT default for English->Serbo-croatian

https://gerrit.wikimedia.org/r/632646

Change 632646 merged by jenkins-bot:
[mediawiki/services/cxserver@master] Make Google MT default for English->Serbo-croatian

https://gerrit.wikimedia.org/r/632646

Change 632838 had a related patch set uploaded (by KartikMistry; owner: KartikMistry):
[operations/deployment-charts@master] Update cxserver to 2020-10-08-053343-production

https://gerrit.wikimedia.org/r/632838

Change 632838 merged by jenkins-bot:
[operations/deployment-charts@master] Update cxserver to 2020-10-08-053343-production

https://gerrit.wikimedia.org/r/632838