[Wikimedia-l] Re: Translating Wikipedia articles

2021-12-19 Thread Ziko van Dijk
Thanks, Vermont, for the explanation! It is indeed good to notice that
„good article“ is a specific term in English Wikipedia, but that other wiki
communities define it differently. E.g., German WP calls these articles
„articles worth to read“ (lesenswerte Artikel).
For the Klexikon, we consider all of our articles to be „good articles“
meaning that they are easy to read, well structured and well composed. (If
they aren‘t, they should be still in the draft name space.)
Kind regards,
Ziko

Chris Gates via Wikimedia-l  schrieb am
Fr. 17. Dez. 2021 um 16:31:

> Hi Ziko,
>
> Content on the Simple English Wikipedia is indeed very uneven. Though most
> of the project’s articles are stubs with not many issues, there are a lot
> of mid-size articles with sourcing and/or accuracy problems.
>
> And given the nature of Simple English, our good articles are not really
> written in the same way that a good article in another language would
> necessarily be written, meaning that machine translation would likely
> require a lot more copyediting than usual to bring it to an acceptable
> level. For specifics, see:
>
> https://simple.wikipedia.org/wiki/Wikipedia:How_to_write_Simple_English_pages#Simple_sentence_structure
>
> Best regards,
> Vermont
>
>
> On Fri, Dec 17, 2021 at 03:55 Ziko van Dijk  wrote:
>
>> Hello James,
>> Interesting, what extra tools you have there.
>> Andreas, in theory Simple English Wikipedia would be great as an
>> international platform for the dissemination of articles. But in
>> reality I have the impression that the content of S.E.WP is very
>> uneven, the quality is very diverse.
>> Galder, nice to hear of these initiatives and what is all possible in
>> a joint action.
>> Kind regards
>> Ziko
>>
>> Am Do., 16. Dez. 2021 um 19:59 Uhr schrieb James Heilman <
>> jmh...@gmail.com>:
>> >
>> > Hey Ziko
>> >
>> > We have moved our medical translation efforts, such that MDWiki is our
>> starting point.
>> >
>> > https://mdwiki.org/wiki/WikiProjectMed:Translation_task_force
>> >
>> > 1) We have also encountered a reference template issue with CTX but
>> have built a work around. Our issue appears to relate to 
>> and the same reference being used multiple times. What we do is we simply
>> expand all the metadata for each instance of the reference before feeding
>> the text into CTX. And then we have a bot that shortens all the instances
>> of a reference back to one.
>> >
>> > 2) One of the benefits of using MDWiki is it allows us to keep
>> references in the lead and use language that is easier to understand, but
>> not be forced to use language as easy as Simple English. It has also
>> allowed us to automatically generate a leaderboard to track progress and
>> impact of our translation efforts. Additionally we only encourage people to
>> translate the leads as that is only that has been medically checked for
>> accuracy.
>> https://mdwiki.toolforge.org/Translation_Dashboard/leaderboard.php
>> >
>> > Best
>> > James
>> >
>> >
>> >
>> > On Thu, Dec 16, 2021 at 5:34 AM Andreas Kolbe 
>> wrote:
>> >>
>> >> Thanks, Ziko. Does anyone use Simple English Wikipedia as a basis for
>> their translations?
>> >>
>> >> I reckon DeepL – which is by far the best machine translation program
>> around, in the languages it covers – might do an even better job with those
>> (provided the Simple English article is itself of good quality, and worth
>> translating).
>> >>
>> >> Andreas
>> >> ___
>> >> Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org,
>> guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines
>> and https://meta.wikimedia.org/wiki/Wikimedia-l
>> >> Public archives at
>> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/R7UQ4EUVVRFXYIYBQB6S25Y7SQQCIB2J/
>> >> To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
>> >
>> >
>> >
>> > --
>> > James Heilman
>> > MD, CCFP-EM, Wikipedian
>> > ___
>> > Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org,
>> guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines
>> and https://meta.wikimedia.org/wiki/Wikimedia-l
>> > Public archives at
>> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/I4QSZFVMBRVU7SBRYROJUMSR5RMQWYHA/
>> > To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
>> ___
>> Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines
>> at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and
>> https://meta.wikimedia.org/wiki/Wikimedia-l
>>
> Public archives at
>> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/NUTGNPULDEVOWA3KL4H5ZCHZKEH7H25M/
>> To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
>
> --
>
> 
> User:Vermont  on 

[Wikimedia-l] Re: Translating Wikipedia articles

2021-12-17 Thread Chris Gates via Wikimedia-l
Hi Ziko,

Content on the Simple English Wikipedia is indeed very uneven. Though most
of the project’s articles are stubs with not many issues, there are a lot
of mid-size articles with sourcing and/or accuracy problems.

And given the nature of Simple English, our good articles are not really
written in the same way that a good article in another language would
necessarily be written, meaning that machine translation would likely
require a lot more copyediting than usual to bring it to an acceptable
level. For specifics, see:
https://simple.wikipedia.org/wiki/Wikipedia:How_to_write_Simple_English_pages#Simple_sentence_structure

Best regards,
Vermont


On Fri, Dec 17, 2021 at 03:55 Ziko van Dijk  wrote:

> Hello James,
> Interesting, what extra tools you have there.
> Andreas, in theory Simple English Wikipedia would be great as an
> international platform for the dissemination of articles. But in
> reality I have the impression that the content of S.E.WP is very
> uneven, the quality is very diverse.
> Galder, nice to hear of these initiatives and what is all possible in
> a joint action.
> Kind regards
> Ziko
>
> Am Do., 16. Dez. 2021 um 19:59 Uhr schrieb James Heilman  >:
> >
> > Hey Ziko
> >
> > We have moved our medical translation efforts, such that MDWiki is our
> starting point.
> >
> > https://mdwiki.org/wiki/WikiProjectMed:Translation_task_force
> >
> > 1) We have also encountered a reference template issue with CTX but have
> built a work around. Our issue appears to relate to  and the
> same reference being used multiple times. What we do is we simply expand
> all the metadata for each instance of the reference before feeding the text
> into CTX. And then we have a bot that shortens all the instances of a
> reference back to one.
> >
> > 2) One of the benefits of using MDWiki is it allows us to keep
> references in the lead and use language that is easier to understand, but
> not be forced to use language as easy as Simple English. It has also
> allowed us to automatically generate a leaderboard to track progress and
> impact of our translation efforts. Additionally we only encourage people to
> translate the leads as that is only that has been medically checked for
> accuracy.
> https://mdwiki.toolforge.org/Translation_Dashboard/leaderboard.php
> >
> > Best
> > James
> >
> >
> >
> > On Thu, Dec 16, 2021 at 5:34 AM Andreas Kolbe 
> wrote:
> >>
> >> Thanks, Ziko. Does anyone use Simple English Wikipedia as a basis for
> their translations?
> >>
> >> I reckon DeepL – which is by far the best machine translation program
> around, in the languages it covers – might do an even better job with those
> (provided the Simple English article is itself of good quality, and worth
> translating).
> >>
> >> Andreas
> >> ___
> >> Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org,
> guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines
> and https://meta.wikimedia.org/wiki/Wikimedia-l
> >> Public archives at
> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/R7UQ4EUVVRFXYIYBQB6S25Y7SQQCIB2J/
> >> To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
> >
> >
> >
> > --
> > James Heilman
> > MD, CCFP-EM, Wikipedian
> > ___
> > Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines
> at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and
> https://meta.wikimedia.org/wiki/Wikimedia-l
> > Public archives at
> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/I4QSZFVMBRVU7SBRYROJUMSR5RMQWYHA/
> > To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
> ___
> Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines
> at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and
> https://meta.wikimedia.org/wiki/Wikimedia-l
> Public archives at
> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/NUTGNPULDEVOWA3KL4H5ZCHZKEH7H25M/
> To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org

-- 


User:Vermont  on Wikimedia
projects
they/them/theirs (why pronouns matter
)
___
Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines at: 
https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and 
https://meta.wikimedia.org/wiki/Wikimedia-l
Public archives at 
https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/B5WSUJKIQUP3PPJIGDFAURQM6ORCFVZV/
To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org

[Wikimedia-l] Re: Translating Wikipedia articles

2021-12-17 Thread Anupam Dutta
Hi All,

For translation from English to Bengali, Google Translation does great jobs
most of the time.
But sadly it has not yet learnt all the basic numbers ..

Regards

Anupamdutta73



শুক্র, ১৭ ডিসেম্বর, ২০২১ তারিখে ৪:৩৩ PM টায় তারিখে Željko Blaće <
zbl...@mi2.hr> লিখেছেন:

> Hi All - it is an interesting discussion to have and compare experiences.
>
> For smaller languages the situation is much worse both on the level of
> technology (DeepL covers only a few languages, GoogleTranslate gets worse)
> and communities (having less human resources to correct on wikis and train
> software off wiki).
>
> I tried to propose Simple as proxy Wiki for CEE Spring 2021 translation
> competition, but it was suboptimal on many levels.
> If anyone wants to share notes on these topics and consider options it
> would be nice to see how to do it off list.
> Maybe on Wikimedia Chat
> https://chat.wmcloud.org/wikimedia/channels/lost-in-translation
>
> Best Z.
>
> On Fri, Dec 17, 2021 at 9:56 AM Ziko van Dijk  wrote:
>
>> Hello James,
>> Interesting, what extra tools you have there.
>> Andreas, in theory Simple English Wikipedia would be great as an
>> international platform for the dissemination of articles. But in
>> reality I have the impression that the content of S.E.WP is very
>> uneven, the quality is very diverse.
>> Galder, nice to hear of these initiatives and what is all possible in
>> a joint action.
>> Kind regards
>> Ziko
>>
>> Am Do., 16. Dez. 2021 um 19:59 Uhr schrieb James Heilman <
>> jmh...@gmail.com>:
>> >
>> > Hey Ziko
>> >
>> > We have moved our medical translation efforts, such that MDWiki is our
>> starting point.
>> >
>> > https://mdwiki.org/wiki/WikiProjectMed:Translation_task_force
>> >
>> > 1) We have also encountered a reference template issue with CTX but
>> have built a work around. Our issue appears to relate to 
>> and the same reference being used multiple times. What we do is we simply
>> expand all the metadata for each instance of the reference before feeding
>> the text into CTX. And then we have a bot that shortens all the instances
>> of a reference back to one.
>> >
>> > 2) One of the benefits of using MDWiki is it allows us to keep
>> references in the lead and use language that is easier to understand, but
>> not be forced to use language as easy as Simple English. It has also
>> allowed us to automatically generate a leaderboard to track progress and
>> impact of our translation efforts. Additionally we only encourage people to
>> translate the leads as that is only that has been medically checked for
>> accuracy.
>> https://mdwiki.toolforge.org/Translation_Dashboard/leaderboard.php
>> >
>> > Best
>> > James
>> >
>> >
>> >
>> > On Thu, Dec 16, 2021 at 5:34 AM Andreas Kolbe 
>> wrote:
>> >>
>> >> Thanks, Ziko. Does anyone use Simple English Wikipedia as a basis for
>> their translations?
>> >>
>> >> I reckon DeepL – which is by far the best machine translation program
>> around, in the languages it covers – might do an even better job with those
>> (provided the Simple English article is itself of good quality, and worth
>> translating).
>> >>
>> >> Andreas
>> >> ___
>> >> Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org,
>> guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines
>> and https://meta.wikimedia.org/wiki/Wikimedia-l
>> >> Public archives at
>> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/R7UQ4EUVVRFXYIYBQB6S25Y7SQQCIB2J/
>> >> To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
>> >
>> >
>> >
>> > --
>> > James Heilman
>> > MD, CCFP-EM, Wikipedian
>> > ___
>> > Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org,
>> guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines
>> and https://meta.wikimedia.org/wiki/Wikimedia-l
>> > Public archives at
>> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/I4QSZFVMBRVU7SBRYROJUMSR5RMQWYHA/
>> > To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
>> ___
>> Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines
>> at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and
>> https://meta.wikimedia.org/wiki/Wikimedia-l
>> Public archives at
>> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/NUTGNPULDEVOWA3KL4H5ZCHZKEH7H25M/
>> To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
>
> ___
> Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines
> at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and
> https://meta.wikimedia.org/wiki/Wikimedia-l
> Public archives at
> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/EAOZBMU6HMJIUW62IUQBRIVWS7VDDBPO/
> To 

[Wikimedia-l] Re: Translating Wikipedia articles

2021-12-17 Thread Željko Blaće
Hi All - it is an interesting discussion to have and compare experiences.

For smaller languages the situation is much worse both on the level of
technology (DeepL covers only a few languages, GoogleTranslate gets worse)
and communities (having less human resources to correct on wikis and train
software off wiki).

I tried to propose Simple as proxy Wiki for CEE Spring 2021 translation
competition, but it was suboptimal on many levels.
If anyone wants to share notes on these topics and consider options it
would be nice to see how to do it off list.
Maybe on Wikimedia Chat
https://chat.wmcloud.org/wikimedia/channels/lost-in-translation

Best Z.

On Fri, Dec 17, 2021 at 9:56 AM Ziko van Dijk  wrote:

> Hello James,
> Interesting, what extra tools you have there.
> Andreas, in theory Simple English Wikipedia would be great as an
> international platform for the dissemination of articles. But in
> reality I have the impression that the content of S.E.WP is very
> uneven, the quality is very diverse.
> Galder, nice to hear of these initiatives and what is all possible in
> a joint action.
> Kind regards
> Ziko
>
> Am Do., 16. Dez. 2021 um 19:59 Uhr schrieb James Heilman  >:
> >
> > Hey Ziko
> >
> > We have moved our medical translation efforts, such that MDWiki is our
> starting point.
> >
> > https://mdwiki.org/wiki/WikiProjectMed:Translation_task_force
> >
> > 1) We have also encountered a reference template issue with CTX but have
> built a work around. Our issue appears to relate to  and the
> same reference being used multiple times. What we do is we simply expand
> all the metadata for each instance of the reference before feeding the text
> into CTX. And then we have a bot that shortens all the instances of a
> reference back to one.
> >
> > 2) One of the benefits of using MDWiki is it allows us to keep
> references in the lead and use language that is easier to understand, but
> not be forced to use language as easy as Simple English. It has also
> allowed us to automatically generate a leaderboard to track progress and
> impact of our translation efforts. Additionally we only encourage people to
> translate the leads as that is only that has been medically checked for
> accuracy.
> https://mdwiki.toolforge.org/Translation_Dashboard/leaderboard.php
> >
> > Best
> > James
> >
> >
> >
> > On Thu, Dec 16, 2021 at 5:34 AM Andreas Kolbe 
> wrote:
> >>
> >> Thanks, Ziko. Does anyone use Simple English Wikipedia as a basis for
> their translations?
> >>
> >> I reckon DeepL – which is by far the best machine translation program
> around, in the languages it covers – might do an even better job with those
> (provided the Simple English article is itself of good quality, and worth
> translating).
> >>
> >> Andreas
> >> ___
> >> Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org,
> guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines
> and https://meta.wikimedia.org/wiki/Wikimedia-l
> >> Public archives at
> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/R7UQ4EUVVRFXYIYBQB6S25Y7SQQCIB2J/
> >> To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
> >
> >
> >
> > --
> > James Heilman
> > MD, CCFP-EM, Wikipedian
> > ___
> > Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines
> at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and
> https://meta.wikimedia.org/wiki/Wikimedia-l
> > Public archives at
> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/I4QSZFVMBRVU7SBRYROJUMSR5RMQWYHA/
> > To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
> ___
> Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines
> at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and
> https://meta.wikimedia.org/wiki/Wikimedia-l
> Public archives at
> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/NUTGNPULDEVOWA3KL4H5ZCHZKEH7H25M/
> To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
___
Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines at: 
https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and 
https://meta.wikimedia.org/wiki/Wikimedia-l
Public archives at 
https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/EAOZBMU6HMJIUW62IUQBRIVWS7VDDBPO/
To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org

[Wikimedia-l] Re: Translating Wikipedia articles

2021-12-17 Thread Ziko van Dijk
Hello James,
Interesting, what extra tools you have there.
Andreas, in theory Simple English Wikipedia would be great as an
international platform for the dissemination of articles. But in
reality I have the impression that the content of S.E.WP is very
uneven, the quality is very diverse.
Galder, nice to hear of these initiatives and what is all possible in
a joint action.
Kind regards
Ziko

Am Do., 16. Dez. 2021 um 19:59 Uhr schrieb James Heilman :
>
> Hey Ziko
>
> We have moved our medical translation efforts, such that MDWiki is our 
> starting point.
>
> https://mdwiki.org/wiki/WikiProjectMed:Translation_task_force
>
> 1) We have also encountered a reference template issue with CTX but have 
> built a work around. Our issue appears to relate to  and the 
> same reference being used multiple times. What we do is we simply expand all 
> the metadata for each instance of the reference before feeding the text into 
> CTX. And then we have a bot that shortens all the instances of a reference 
> back to one.
>
> 2) One of the benefits of using MDWiki is it allows us to keep references in 
> the lead and use language that is easier to understand, but not be forced to 
> use language as easy as Simple English. It has also allowed us to 
> automatically generate a leaderboard to track progress and impact of our 
> translation efforts. Additionally we only encourage people to translate the 
> leads as that is only that has been medically checked for accuracy. 
> https://mdwiki.toolforge.org/Translation_Dashboard/leaderboard.php
>
> Best
> James
>
>
>
> On Thu, Dec 16, 2021 at 5:34 AM Andreas Kolbe  wrote:
>>
>> Thanks, Ziko. Does anyone use Simple English Wikipedia as a basis for their 
>> translations?
>>
>> I reckon DeepL – which is by far the best machine translation program 
>> around, in the languages it covers – might do an even better job with those 
>> (provided the Simple English article is itself of good quality, and worth 
>> translating).
>>
>> Andreas
>> ___
>> Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines at: 
>> https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and 
>> https://meta.wikimedia.org/wiki/Wikimedia-l
>> Public archives at 
>> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/R7UQ4EUVVRFXYIYBQB6S25Y7SQQCIB2J/
>> To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
>
>
>
> --
> James Heilman
> MD, CCFP-EM, Wikipedian
> ___
> Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines at: 
> https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and 
> https://meta.wikimedia.org/wiki/Wikimedia-l
> Public archives at 
> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/I4QSZFVMBRVU7SBRYROJUMSR5RMQWYHA/
> To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org
___
Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines at: 
https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and 
https://meta.wikimedia.org/wiki/Wikimedia-l
Public archives at 
https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/NUTGNPULDEVOWA3KL4H5ZCHZKEH7H25M/
To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org

[Wikimedia-l] Re: Translating Wikipedia articles

2021-12-16 Thread James Heilman
Hey Ziko

We have moved our medical translation efforts, such that MDWiki is our
starting point.

https://mdwiki.org/wiki/WikiProjectMed:Translation_task_force

1) We have also encountered a reference template issue with CTX but have
built a work around. Our issue appears to relate to  and the
same reference being used multiple times. What we do is we simply expand
all the metadata for each instance of the reference before feeding the text
into CTX. And then we have a bot that shortens all the instances of a
reference back to one.

2) One of the benefits of using MDWiki is it allows us to keep references
in the lead and use language that is easier to understand, but not be
forced to use language as easy as Simple English. It has also allowed us to
automatically generate a leaderboard to track progress and impact of our
translation efforts. Additionally we only encourage people to translate the
leads as that is only that has been medically checked for accuracy.
https://mdwiki.toolforge.org/Translation_Dashboard/leaderboard.php

Best
James



On Thu, Dec 16, 2021 at 5:34 AM Andreas Kolbe  wrote:

> Thanks, Ziko. Does anyone use Simple English Wikipedia as a basis for
> their translations?
>
> I reckon DeepL – which is by far the best machine translation program
> around, in the languages it covers – might do an even better job with those
> (provided the Simple English article is itself of good quality, and worth
> translating).
>
> Andreas
> ___
> Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines
> at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and
> https://meta.wikimedia.org/wiki/Wikimedia-l
> Public archives at
> https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/R7UQ4EUVVRFXYIYBQB6S25Y7SQQCIB2J/
> To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org



-- 
James Heilman
MD, CCFP-EM, Wikipedian
___
Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines at: 
https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and 
https://meta.wikimedia.org/wiki/Wikimedia-l
Public archives at 
https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/I4QSZFVMBRVU7SBRYROJUMSR5RMQWYHA/
To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org

[Wikimedia-l] Re: Translating Wikipedia articles

2021-12-16 Thread Andreas Kolbe
Thanks, Ziko. Does anyone use Simple English Wikipedia as a basis for their
translations?

I reckon DeepL – which is by far the best machine translation program
around, in the languages it covers – might do an even better job with those
(provided the Simple English article is itself of good quality, and worth
translating).

Andreas
___
Wikimedia-l mailing list -- wikimedia-l@lists.wikimedia.org, guidelines at: 
https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and 
https://meta.wikimedia.org/wiki/Wikimedia-l
Public archives at 
https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org/message/R7UQ4EUVVRFXYIYBQB6S25Y7SQQCIB2J/
To unsubscribe send an email to wikimedia-l-le...@lists.wikimedia.org

[Wikimedia-l] Re: Translating Wikipedia articles

2021-12-16 Thread Galder Gonzalez Larrañaga
Thanks Ziko for your point of view,
For a small Wikipedia like the Basque Wikipedia ContentTranslation is a game 
changer. Our partners from Elhuyar Foundation have been working with WMF to 
provide their machine translation system (Elia) so, now, we can translate 
between eu, es, ca, gl, fr and en. This makes everything faster, and, although 
the translation is far from perfect, it saves a lot of time. Some issues you 
are presenting, especially those related to templates and references, should be 
worked better. As far as I know, the system uses template links from Wikidata, 
and you can add which labels are which in the other languages, using alias in 
the templatedata.

The system may be improved, but the product is way better than we had some 
years ago.

Galder

From: Ziko van Dijk 
Sent: Thursday, December 16, 2021 12:40 PM
To: Wikimedia Mailing List 
Subject: [Wikimedia-l] Translating Wikipedia articles

Hello,

I would like to share some experiences with the Content Translation
tool when I translated an article from German to English Wikipedia.
There are issues that could need a movement wide discussion,
especially: the use of references, and the use of automatic
translation.

Links:
About the tool: https://www.mediawiki.org/wiki/Content_translation
My article: https://en.wikipedia.org/wiki/Autumn_Crisis_1850 (The
subject is a conflict in the German history of the 19th century.)

1. Why use the tool at all?
In general, I am quite impressed with the tool as it is now. It saves
me the hassle to deal with Wikidata, as the link is made
automatically. It also makes it easy to use the pictures in the
original article. Sometimes, the tool even manages to translate an
internal link to another Wikipedia article. (In many other cases, I
had to do the linking manually, as "Otto von Manteuffel" was not
recognised as the same person as in the already existing article
[[en:Otto Theodor von Manteuffel]].)
I also like that the tool saves my edits, so that I can continue
translating another day without having to save the text externally.

2. What about automatic translation?
I decided to make use of automatic translation, name "deepl.com" as
this website does an excellent job (at least between the languages
that I understand). The main reason for me is to save time: the
website knows many English terms I would have to look up. It is not
just about the terms but also the correct prepositions etc. Also: is
it "campsite" or "camp site"?

The automatic translation, still, is not perfect, and I would never
advise to use it without checking each and every sentence. You always
have to read carefully the original paragraph and then the proposed
translation. At that occasion, I consult my online dictionary a lot.

And, frankly, when I translate from a foreign language (such English)
to my native language (German), I don't use the automatic translation
but create the German text all by myself. My own wording may differ
significantly from the original, because it is my goal to create a
readable German article, not to preserve the original text with all
its details and difficulties.

Usual problems with the automatic translation of deepl.com are:
* a strange wording, even the omission of whole words
* a misunderstanding of the original text; for example, the original
German article in this case dealt with the "Confederation" and the
"Union" in Germany 1850, and deepl.com at one occasion wrote
"Confederation" where it should have been "Union"
* Deepl.com does not always recognize proper scientific terms. Also,
in some cases, the German term differs from the usual one in English,
e.g. the "German Dualism" is usually called the "Austrian-Prussian
rivalry" in English.
* Deepl.com sometimes translated "Kurhessen" to "Kurhessen", in other
cases to "Electoral Hesse", in others to "Electorate of Hesse". All of
these translations are correct, but I decided to use only "Kurhessen"
in the final text.

So in general, I think that the translations from deepl.com are often
astonishingly great. But you have to check them carefully.

3. Should we integrate automatic translation into the tool?
As you know, many Wikipedia language versions do not allow for the use
of automatic translation within our tool. The main reason: some people
deliver articles without a clean-up. The result are unreadable or
misleading new articles in, e.g., English Wikipedia. (In German
Wikipedia, we call unreadable articles "Babel Fish accidents".)

But should the discussion end here? For example, we might want to
allow automatic translation for those editors who actually do the
clean-up. Why not giving the permission to editors who apply for the
right to use automatic translation within the tool? Should an editor
indeed deliver bad translations, the editor could still loose this
right.

By the way, I made use of deepl.com translations by simply
copying+pasting the paragraphs. Alas, I had to do this for every
single paragraph which is