Status Update about effort to translate the FreeBSD Documentation from en_US to pt_BR

[Edson Brandi]

Hello Guys,

First of all, i'm sorry for the text wall... But I think it's worth
sharing with you what we did in the last 2 weeks, because I believe
that what we have learned can be useful to other translation groups.

About two weeks ago Danilo (dbaio@) mentioned here in the list about
the tests he was doing with the Zanata translation platform, and the
.po files that can be generated with the docproj toolset.

Talking to him in private we concluded that we should try to resume
the project to translate the FreeBSD documentation into our language
using these tools to simplify the translation workflow. And since we
made this decision I can say that the last 2 weeks have been extremely
productive for us.

Before starting the project itself, we have prepared the necessary
infrastructure to receive the translations, than we re-enabled the
github repository maintained by the Brazilian FreeBSD user group
(FUG-BR) which was used in the past for the translation initiatives,
and it was loaded with .po files generated with docproj toolset from
all the books and articles available in the en_US.ISO8859-1 directory,
and we have also created a project on the Zanata platform with them.

We also configured an automated pipeline with jenkins to build the
documents from the translated .po files downloaded Zanata platform,
and we've published a new website with basic instructions on how
people could contribute to the translation,

Once everything was ready, we went out looking for volunteers to help
us with the translation.

When we presented the new simplified translation process to our
community we had an excellent acceptance, which allowed us to quickly
reach the number of 22 volunteers working on the translation and
revision of the documents.

We talked to many of the volunteers to understand what method they
would normally use to do their translations and found that most of
them would use google translator to draft an initial translation, and
than this initial draft goes through a meticulously review, to ensure
that the text are in the correct context.

Based on this finding, we decided to make the work of the volunteers
even simpler and one of our volunteers builded a NLTK script that
allowed us to extract the text contained in the msgid tags of an .po
files and to translate them using Google Cloud Translation API,
generating a new .po file that already have an initial translation at
msgstr tags. T

This simple procedure avoided a lot of manual work for our volunteers
and have speedup the translation process a lot.

After these initial 2 weeks of work, we have the following results to share:

* 100% of the content of all files (7 books and 35 articles) has
already been translated from en_US into pt_BR using Google Cloud
Translation API

* ~42% of the translated content of all files has already been
reviewed by at least one volunteer.

* ~11% of the translated content of all files has already been
reviewed by at least one  second  volunteer, and are considered as
"finished" - based in our previous experience if a document has passed
successfully by at least 2 different revisions made by 2 different
volunteers, it´s translation can be considered finished.

While there is still a lot work to be done, the progress we have made
in these 2 weeks has been enough to keep our volunteers motivated to
keep moving forward :)

Than i'm sure that very soon we will have some documents ready to be
committed to our SVN repository.

For those who wish to track our progress, i think that following urls
can be useful:

Preview of pt_BR builded documents:
https://doc.fug.com.br/jenkins/

Our project website:
https://doc.fug.com.br

Our project at Zanata Platform:
https://translate.zanata.org/project/view/FreeBSD

Our github repository:
https://github.com/doc-br/zanata

-- 
[  ]'s Brandi
--
Edson Brandi
ebrandi at FreeBSD.org

OpenPGP Key: 0x209E9B54FFD3035B
443B 5363 564F 06C3 EA54  9482 209E 9B54 FFD3 035B