machine translation

List overview All Threads
Download

newer

older

Recognition of the Wikivoyage...

Wikimedia District of Columbia:...

Amir E. Aharoni

2 May 2017 2 May '17

5:07 p.m.

2017-05-02 18:20 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>:

...

Brute force solution; turn the ContentTranslation off. Really stupid solution.

... Then I guess you don't mind that I'm changing the thread name :)

...

The next solution; turn the Yandex engine off. That would solve a part of the problem. Kind of lousy solution though.

...

What about adding a language model that warns when the language constructs gets to weird? It is like a "test" for the translation. The CT is used for creating a translation, but the language model is used for verifying if the translation is good enough. If it does not validate against the language model it should simply not be published to the main name space. It will still be possible to create a draft, but then the user is completely aware that the translation isn't good enough. Such a language model should be available as a test for any article, as it can be used as a quality measure for the article. It is really a quantity measure for the well-spokenness of the article, but that isn't quite so intuitive.

So, I'll allow myself to guess that you are talking about one particular language, probably Norwegian. Several technical facts: 1. In the past there were several cases in which translators to different languages who reported common translation mistakes to me. I passed them on to Yandex developers, with whom I communicate quite regularly. They acknowledged receiving all of them. I am aware of at least one such common mistake that was fixed; possibly there were more. If you can give me a list of such mistakes for Norwegian, I'll be very happy to pass them on. I absolutely cannot promise that they will be fixed upstream, but it's possible. 2. In Norwegian, Apertium is used for translating between the two varieties of Norwegian itself (Bokmål and Nynorsk), and from other Scandinavian languages. That's probably why it works so well—they are similar in grammar, vocabulary, and narrative style (I'll pass it on to Apertium developers—I'm sure they'll be happy to hear it). Unfortunately, machine translation from English is not available in Apertium. Apertium works best with very similar languages, and English has two characteristics, which are unfortunate when combined: it is both the most popular source for translation into almost all other languages (including Norwegian), and it is not _very_ similar to any other languages (except maybe Scots). Machine translation from English into Norwegian is only possible with Yandex at the moment. More engines may be added in the future, but at the moment that's all we have. That's why disabling Yandex completely would indeed be a lousy solution: A lot of people say that without machine translation integration Content Translation is useless. Not all users think like that, but many do. 3. We can define a numerical threshold of acceptable percentage of machine translation post-editing. Currently it's 75%. It's a tad embarrassing, but it's hard-coded at the moment, but it can be very easily be made into a variable per language. If the translator tries to publish a page in which less than that is modified, a warning will be shown. 4. I'm not sure what do you mean by "language model". If it's any kind of a linguistic engine, then it's definitely not within the resources that the Language team itself can currently dedicate. However, if somebody who knows Norwegian and some programming will write a script that analyzes common bad constructs in a Wikipedia dump, this will be very useful. This would basically be an upgraded version of suggestion #1 above. (In my spare time as a volunteer I'm doing something comparable for Hebrew, although not for translation, but for improving how MediaWiki link trails work.)

Show replies by date

John Erling Blad

2 May 2 May

6:47 p.m.

Actually this _is_ about turning ContentTranslation off, that is what several users in the community want. They block people using the extension and delete the translated articles. Use of ContentTranslation has become a rather contentious case. Yandex as a general translation engine to be able to read some alien language is quite good, but as an engine to produce written text it is not very good at all. In fact it often creates quite horrible Norwegian, even for closely related languages. One quite common problem is reordering of words into meaningless constructs, an other problem is reordering lexical gender in weird ways. The English preposition "a" is often translated as "en" in a propositional phrase, and then the gender is added to the following phrase. That gives a translation of "Oppland is a county in…" into something like "Oppland er en fylket i…" This should be "Oppland er et fylke i…". (I just checked and it seems like Yandex messes up a lot less now than previously, but it is still pretty bad.) Apertium works because the language is closely related, Yandex does not work because it is used between very different languages. People try to use Yandex and gets disappointed, and falsely conclude that all language translations are equally weird. They are not, but Yandex translations are weird. The numerical threshold does not work. The reason is simple, the number of fixes depends on language constructs that fails, and that is simply not a constant for small text fragments. Perhaps if we could flag specific language constructs that is known to give a high percentage of failures, and if the translator must check those sentences. One such language construct is disappearances between the preposition and the gender of the following term in a prepositional phrase. If they are not similar, then the sentence must be checked. It is not always wrong to write "en jenta" in Norwegian, but it is likely to be wrong. A language model could be a statistical model for the language itself, not for the translation into that language. We don't want a perfect language model, but a sufficient language model to mark weird constructs. A very simple solution could simply be to mark tri-grams that does not already exist in the text base for the destination as possible errors. It is not necessary to do a live check, but at least do it before the page can be saved. Note the difference in what Yandex do and what we want to achieve; Yandex translates a text between two different languages, without any clear reason why. It is not to important if there are weird constructs in the text, as long as it is usable in "some" context. We translate a text for the purpose of republishing it. The text should be usable and easily readable in that language. On Tue, May 2, 2017 at 7:07 PM, Amir E. Aharoni < amir.aharoni(a)mail.huji.ac.il> wrote:

...

2017-05-02 18:20 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>:

Brute force solution; turn the ContentTranslation off. Really stupid solution.

... Then I guess you don't mind that I'm changing the thread name :)

The next solution; turn the Yandex engine off. That would solve a part of the problem. Kind of lousy solution though.

What about adding a language model that warns when the language

constructs

gets to weird? It is like a "test" for the translation. The CT is used

for

creating a translation, but the language model is used for verifying if

the

translation is good enough. If it does not validate against the language model it should simply not be published to the main name space. It will still be possible to create a draft, but then the user is completely

aware

that the translation isn't good enough. Such a language model should be available as a test for any article, as

can be used as a quality measure for the article. It is really a quantity measure for the well-spokenness of the article, but that isn't quite so intuitive.

Amir E. Aharoni

7:25 p.m.

2017-05-02 21:47 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>:

...

Yandex as a general translation engine to be able to read some alien language is quite good, but as an engine to produce written text it is not very good at all.

... Nor is it supposed to be. A translator is a person. Machine translation software is not a person, it's software. It's a tool that is supposed to help a human translator produce a good written text more quickly. If it doesn't make this work faster, it can and should be disabled. If no translator

...

In fact it often creates quite horrible Norwegian, even for closely related languages. One quite common problem is reordering of words into meaningless constructs, an other problem is reordering lexical gender in weird ways. The English preposition "a" is often translated as "en" in a propositional phrase, and then the gender is added to the following phrase. That gives a translation of "Oppland is a county in…" into something like "Oppland er en fylket i…" This should be "Oppland er et fylke i…".

I suggest making a page with a list of such examples, so that the machine translation developers could read it.

...

(I just checked and it seems like Yandex messes up a lot less now than previously, but it is still pretty bad.)

I guess that this is something that Yandex developers will be happy to hear :) More seriously, it's quite possible that they already used some of the translations made by the Norwegian Wikipedia community. In addition to being published as an article, each translated paragraph is saved into parallel corpora, and machine translation developers read the edited text and use it to improve their software. This is completely open and usable by all machine translation developers, not only for Yandex.

...

The numerical threshold does not work. The reason is simple, the number of fixes depends on language constructs that fails, and that is simply not a constant for small text fragments. Perhaps if we could flag specific language constructs that is known to give a high percentage of failures, and if the translator must check those sentences. One such language construct is disappearances between the preposition and the gender of the following term in a prepositional phrase.

The question is how would we do it with our software. I simply cannot imagine doing it with the current MediaWiki platform, unless we develop a sophisticated NLP engine, although it's possible I'm exaggerating or forgetting something.

...

A language model could be a statistical model for the language itself, not for the translation into that language. We don't want a perfect language model, but a sufficient language model to mark weird constructs. A very simple solution could simply be to mark tri-grams that does not already exist in the text base for the destination as possible errors. It is not necessary to do a live check, but at least do it before the page can be saved.

See above—we don't have support for plugging something like that into our workflow. Perhaps one day some AI/machine-learning system like ORES would be able to do it. Maybe it could be an extension to ORES itself.

...

Note the difference in what Yandex do and what we want to achieve; Yandex translates a text between two different languages, without any clear reason why. It is not to important if there are weird constructs in the text, as long as it is usable in "some" context. We translate a text for the purpose of republishing it. The text should be usable and easily readable in that language.

This is a well-known problem in machine translation: domain. Professional industrial translation powerhouses use internally-customized machine translation engines that specialize on particular domains, such as medicine, law, or news. In theory, it would make a lot of sense to have a customized machine translation engine for encyclopedic articles, or maybe even for several different styles of encyclopedic articles (biography, science, history, etc.). For now what we have is a very general-purpose consumer-oriented engine. I hope it changes in the future.

John Erling Blad

3 May 3 May

10:37 a.m.

...

More seriously, it's quite possible that they already used some of the translations made by the Norwegian Wikipedia community. In addition to being published as an article, each translated paragraph is saved into parallel corpora, and machine translation developers read the edited text and use it to improve their software. This is completely open and usable by all machine translation developers, not only for Yandex.

It is quite possible the Yandex people has done something as the translations are a lot better now than previously. It also imply that it is really important to correct the text inside CT. The question is how would we do it with our software. I simply cannot

...

imagine doing it with the current MediaWiki platform, unless we develop a sophisticated NLP engine, although it's possible I'm exaggerating or forgetting something.

There are several places this can be inserted, both in VE and in MW. What I want is a kind of rather simple language model, but Aharoni proposed Languagetools in private communication. That lib is very interesting. Perhaps one day some AI/machine-learning system like ORES would be able to

...

do it. Maybe it could be an extension to ORES itself.

I've seen language models implemented as neural nets, but it is not necessary to do it like that. Actually it is more common to do it with plain statistics. On Tue, May 2, 2017 at 9:25 PM, Amir E. Aharoni < amir.aharoni(a)mail.huji.ac.il> wrote: > 2017-05-02 21:47 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>: > > > Yandex as a general translation engine to be able to read some alien > > language is quite good, but as an engine to produce written text it is > not > > very good at all. > > > ... Nor is it supposed to be. > > A translator is a person. Machine translation software is not a person, > it's software. It's a tool that is supposed to help a human translator > produce a good written text more quickly. If it doesn't make this work > faster, it can and should be disabled. If no translator > > > > In fact it often creates quite horrible Norwegian, even > > for closely related languages. One quite common problem is reordering of > > words into meaningless constructs, an other problem is reordering lexical > > gender in weird ways. The English preposition "a" is often translated as > > "en" in a propositional phrase, and then the gender is added to the > > following phrase. That gives a translation of "Oppland is a county in…" > > into something like "Oppland er en fylket i…" This should be "Oppland er > > et fylke i…". > > > > I suggest making a page with a list of such examples, so that the machine > translation developers could read it. > > > > (I just checked and it seems like Yandex messes up a lot less now than > > previously, but it is still pretty bad.) > > > > I guess that this is something that Yandex developers will be happy to hear > :)

...

> > > > > The numerical threshold does not work. The reason is simple, the number > of > > fixes depends on language constructs that fails, and that is simply not a > > constant for small text fragments. Perhaps if we could flag specific > > language constructs that is known to give a high percentage of failures, > > and if the translator must check those sentences. One such language > > construct is disappearances between the preposition and the gender of the > > following term in a prepositional phrase. > > > > The question is how would we do it with our software. I simply cannot

...

imagine doing it with the current MediaWiki platform, unless we develop a sophisticated NLP engine, although it's possible I'm exaggerating or forgetting something.

> > > > A language model could be a statistical model for the language itself, > not > > for the translation into that language. We don't want a perfect language > > model, but a sufficient language model to mark weird constructs. A very > > simple solution could simply be to mark tri-grams that does not already > > exist in the text base for the destination as possible errors. It is not > > necessary to do a live check, but at least do it before the page can be > > saved. > > > > See above—we don't have support for plugging something like that into our > workflow. > > Perhaps one day some AI/machine-learning system like ORES would be able to

...

do it. Maybe it could be an extension to ORES itself.

> > > > Note the difference in what Yandex do and what we want to achieve; Yandex > > translates a text between two different languages, without any clear > reason > > why. It is not to important if there are weird constructs in the text, as > > long as it is usable in "some" context. We translate a text for the > purpose > > of republishing it. The text should be usable and easily readable in that > > language. > > > > This is a well-known problem in machine translation: domain. > > Professional industrial translation powerhouses use internally-customized > machine translation engines that specialize on particular domains, such as > medicine, law, or news. In theory, it would make a lot of sense to have a > customized machine translation engine for encyclopedic articles, or maybe > even for several different styles of encyclopedic articles (biography, > science, history, etc.). For now what we have is a very general-purpose > consumer-oriented engine. I hope it changes in the future. > _______________________________________________ > Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ > wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ > wiki/Wikimedia-l > New messages to: Wikimedia-l(a)lists.wikimedia.org > Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, > <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> >

Amir E. Aharoni

10:50 a.m.

[ Meta-comment: We usually call it "CX" and not "CT".[1] ] 2017-05-03 13:37 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>:

...

all machine translation developers, not only for Yandex.

It is quite possible the Yandex people has done something as the translations are a lot better now than previously. It also imply that it is really important to correct the text inside CT.

Absolutely. All CX users must be encouraged to do this. Translation is done by humans. That's the whole point. Content Translation is not a machine translation tool. It's an article creation tool, which includes optional machine translation for some language pairs. The Content Translation user interface has three warning messages that discourage publishing unedited machine translation,[2][3][4] and several of CX FAQs address this as well.[1] If a user publishes an unedited machine translation, it should be handled just like any other problematic page: it must be edited, moved to a draft, or deleted, and the creating user should be warned. [1] https://www.mediawiki.org/wiki/Content_translation/Documentation/FAQ [2] https://translatewiki.net/w/i.php?title=Special:Translations&message=Me… [3] https://translatewiki.net/w/i.php?title=Special:Translations&message=Me… [4] https://translatewiki.net/w/i.php?title=Special:Translations&message=Me… -- Amir Elisha Aharoni · אָמִיר אֱלִישָׁע אַהֲרוֹנִי http://aharoni.wordpress.com ‪“We're living in pieces, I want to live in peace.” – T. Moore‬

Bodhisattwa Mandal

2 May 2 May

8:09 p.m.

Content translation with Yandex is also a problem in Bengali Wikipedia. Some users have grown a tendency to create machine translated meaningless articles with this extension to increase edit count and article count. This has increased the workloads of admins to find and delete those articles. Yandex is not ready for many languages and it is better to shut it. We don't need it in Bengali. Regards On May 3, 2017 12:17 AM, "John Erling Blad" <jeblad(a)gmail.com> wrote:

...

2017-05-02 18:20 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>:

Brute force solution; turn the ContentTranslation off. Really stupid solution.

... Then I guess you don't mind that I'm changing the thread name :)

The next solution; turn the Yandex engine off. That would solve a part of the problem. Kind of lousy solution though.

What about adding a language model that warns when the language

constructs

gets to weird? It is like a "test" for the translation. The CT is used

for

creating a translation, but the language model is used for verifying if

the > translation is good enough. If it does not validate against the

language

model it should simply not be published to the main name space. It will still be possible to create a draft, but then the user is completely

aware

that the translation isn't good enough. Such a language model should be available as a test for any article, as

it > can be used as a quality measure for the article. It is really a

quantity

measure for the well-spokenness of the article, but that isn't quite so intuitive.

to Yandex developers, with whom I communicate quite regularly. They acknowledged receiving all of them. I am aware of at least one such

common

mistake that was fixed; possibly there were more. If you can give me a

list

of such mistakes for Norwegian, I'll be very happy to pass them on. I absolutely cannot promise that they will be fixed upstream, but it's possible. 2. In Norwegian, Apertium is used for translating between the two

varieties

of Norwegian itself (Bokmål and Nynorsk), and from other Scandinavian languages. That's probably why it works so well—they are similar in grammar, vocabulary, and narrative style (I'll pass it on to Apertium developers—I'm sure they'll be happy to hear it). Unfortunately, machine translation from English is not available in Apertium. Apertium works

best

with very similar languages, and English has two characteristics, which

are

unfortunate when combined: it is both the most popular source for translation into almost all other languages (including Norwegian), and it is not _very_ similar to any other languages (except maybe Scots).

Machine

translation from English into Norwegian is only possible with Yandex at

the

moment. More engines may be added in the future, but at the moment that's all we have. That's why disabling Yandex completely would indeed be a

lousy

solution: A lot of people say that without machine translation

integration

Content Translation is useless. Not all users think like that, but many

do.

3. We can define a numerical threshold of acceptable percentage of

machine

translation post-editing. Currently it's 75%. It's a tad embarrassing,

but

it's hard-coded at the moment, but it can be very easily be made into a variable per language. If the translator tries to publish a page in which less than that is modified, a warning will be shown. 4. I'm not sure what do you mean by "language model". If it's any kind

of a

linguistic engine, then it's definitely not within the resources that the Language team itself can currently dedicate. However, if somebody who

knows

Norwegian and some programming will write a script that analyzes common

bad

constructs in a Wikipedia dump, this will be very useful. This would basically be an upgraded version of suggestion #1 above. (In my spare

time

as a volunteer I'm doing something comparable for Hebrew, although not

for

translation, but for improving how MediaWiki link trails work.) _______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ wiki/Wikimedia-l New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

_______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ wiki/Wikimedia-l New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

Pharos

8:16 p.m.

I think it all depends on the level of engagement of the human translator. When the tool is used in the right way, it is a fantastic tool. Maybe we can find better methods to nudge people toward taking their time and really doing work on their translations. Thanks, Pharos On Tue, May 2, 2017 at 4:09 PM, Bodhisattwa Mandal < bodhisattwa.rgkmc(a)gmail.com> wrote:

...

Actually this _is_ about turning ContentTranslation off, that is what several users in the community want. They block people using the

extension

and delete the translated articles. Use of ContentTranslation has become

rather contentious case. Yandex as a general translation engine to be able to read some alien language is quite good, but as an engine to produce written text it is

not

very good at all. In fact it often creates quite horrible Norwegian, even for closely related languages. One quite common problem is reordering of words into meaningless constructs, an other problem is reordering lexical gender in weird ways. The English preposition "a" is often translated as "en" in a propositional phrase, and then the gender is added to the following phrase. That gives a translation of "Oppland is a county in…" into something like "Oppland er en fylket i…" This should be "Oppland er et fylke i…". (I just checked and it seems like Yandex messes up a lot less now than previously, but it is still pretty bad.) Apertium works because the language is closely related, Yandex does not work because it is used between very different languages. People try to

use

Yandex and gets disappointed, and falsely conclude that all language translations are equally weird. They are not, but Yandex translations are weird. The numerical threshold does not work. The reason is simple, the number

fixes depends on language constructs that fails, and that is simply not a constant for small text fragments. Perhaps if we could flag specific language constructs that is known to give a high percentage of failures, and if the translator must check those sentences. One such language construct is disappearances between the preposition and the gender of the following term in a prepositional phrase. If they are not similar, then

the

sentence must be checked. It is not always wrong to write "en jenta" in Norwegian, but it is likely to be wrong. A language model could be a statistical model for the language itself,

not

for the translation into that language. We don't want a perfect language model, but a sufficient language model to mark weird constructs. A very simple solution could simply be to mark tri-grams that does not already exist in the text base for the destination as possible errors. It is not necessary to do a live check, but at least do it before the page can be saved. Note the difference in what Yandex do and what we want to achieve; Yandex translates a text between two different languages, without any clear

reason

why. It is not to important if there are weird constructs in the text, as long as it is usable in "some" context. We translate a text for the

purpose

of republishing it. The text should be usable and easily readable in that language. On Tue, May 2, 2017 at 7:07 PM, Amir E. Aharoni < amir.aharoni(a)mail.huji.ac.il> wrote: > 2017-05-02 18:20 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>: > > > Brute force solution; turn the ContentTranslation off. Really stupid > > solution. > > > ... Then I guess you don't mind that I'm changing the thread name :) > > > > The next solution; turn the Yandex engine off. That would solve a > > part of the problem. Kind of lousy solution though. > > > > > What about adding a language model that warns when the language > constructs > > gets to weird? It is like a "test" for the translation. The CT is

used

> for > > creating a translation, but the language model is used for verifying

the > translation is good enough. If it does not validate against the

language > > model it should simply not be published to the main name space. It

will

> > still be possible to create a draft, but then the user is completely > aware > > that the translation isn't good enough. > > > > Such a language model should be available as a test for any article,

it > can be used as a quality measure for the article. It is really a

quantity > > measure for the well-spokenness of the article, but that isn't quite

> > intuitive. > > > > So, I'll allow myself to guess that you are talking about one

particular

> language, probably Norwegian. > > Several technical facts: > > 1. In the past there were several cases in which translators to

different

languages who reported common translation mistakes to me. I passed them

to Yandex developers, with whom I communicate quite regularly. They acknowledged receiving all of them. I am aware of at least one such

common

mistake that was fixed; possibly there were more. If you can give me a

list

varieties > of Norwegian itself (Bokmål and Nynorsk), and from other Scandinavian > languages. That's probably why it works so well—they are similar in > grammar, vocabulary, and narrative style (I'll pass it on to Apertium > developers—I'm sure they'll be happy to hear it). Unfortunately,

machine

translation from English is not available in Apertium. Apertium works

best

with very similar languages, and English has two characteristics, which

are > unfortunate when combined: it is both the most popular source for > translation into almost all other languages (including Norwegian), and

is not _very_ similar to any other languages (except maybe Scots).

Machine

translation from English into Norwegian is only possible with Yandex at

the > moment. More engines may be added in the future, but at the moment

that's

all we have. That's why disabling Yandex completely would indeed be a

lousy

solution: A lot of people say that without machine translation

integration

Content Translation is useless. Not all users think like that, but many

do.

3. We can define a numerical threshold of acceptable percentage of

machine

translation post-editing. Currently it's 75%. It's a tad embarrassing,

but > it's hard-coded at the moment, but it can be very easily be made into a > variable per language. If the translator tries to publish a page in

which

less than that is modified, a warning will be shown. 4. I'm not sure what do you mean by "language model". If it's any kind

of a > linguistic engine, then it's definitely not within the resources that

the

Language team itself can currently dedicate. However, if somebody who

knows

Norwegian and some programming will write a script that analyzes common

bad

constructs in a Wikipedia dump, this will be very useful. This would basically be an upgraded version of suggestion #1 above. (In my spare

time

as a volunteer I'm doing something comparable for Hebrew, although not

for

Yaroslav Blanter

3 May 3 May

7:33 a.m.

Creating machine translations only in the draft space (or in the user space in the projects which do not have draft) could help. Cheers Yaroslav On Tue, May 2, 2017 at 10:16 PM, Pharos <pharosofalexandria(a)gmail.com> wrote:

...

This

Actually this _is_ about turning ContentTranslation off, that is what several users in the community want. They block people using the

extension > and delete the translated articles. Use of ContentTranslation has

become

rather contentious case. Yandex as a general translation engine to be able to read some alien language is quite good, but as an engine to produce written text it is

not > very good at all. In fact it often creates quite horrible Norwegian,

even

> for closely related languages. One quite common problem is reordering

> words into meaningless constructs, an other problem is reordering

lexical

> gender in weird ways. The English preposition "a" is often translated

> "en" in a propositional phrase, and then the gender is added to the > following phrase. That gives a translation of "Oppland is a county

in…"

> into something like "Oppland er en fylket i…" This should be "Oppland

et fylke i…". (I just checked and it seems like Yandex messes up a lot less now than previously, but it is still pretty bad.) Apertium works because the language is closely related, Yandex does not work because it is used between very different languages. People try to

use > Yandex and gets disappointed, and falsely conclude that all language > translations are equally weird. They are not, but Yandex translations

are

weird. The numerical threshold does not work. The reason is simple, the number

of > fixes depends on language constructs that fails, and that is simply

not a

> constant for small text fragments. Perhaps if we could flag specific > language constructs that is known to give a high percentage of

failures,

> and if the translator must check those sentences. One such language > construct is disappearances between the preposition and the gender of

the

following term in a prepositional phrase. If they are not similar, then

the

sentence must be checked. It is not always wrong to write "en jenta" in Norwegian, but it is likely to be wrong. A language model could be a statistical model for the language itself,

not > for the translation into that language. We don't want a perfect

language

> model, but a sufficient language model to mark weird constructs. A very > simple solution could simply be to mark tri-grams that does not

already

> exist in the text base for the destination as possible errors. It is

not

> necessary to do a live check, but at least do it before the page can

> saved. > > Note the difference in what Yandex do and what we want to achieve;

Yandex

translates a text between two different languages, without any clear

reason > why. It is not to important if there are weird constructs in the text,

long as it is usable in "some" context. We translate a text for the

purpose > of republishing it. The text should be usable and easily readable in

that

> language. > > > > On Tue, May 2, 2017 at 7:07 PM, Amir E. Aharoni < > amir.aharoni(a)mail.huji.ac.il> wrote: > > > 2017-05-02 18:20 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>: > > > > > Brute force solution; turn the ContentTranslation off. Really

stupid

> > solution. > > > ... Then I guess you don't mind that I'm changing the thread name :) > > > > The next solution; turn the Yandex engine off. That would solve a > > part of the problem. Kind of lousy solution though. > > > > > What about adding a language model that warns when the language > constructs > > gets to weird? It is like a "test" for the translation. The CT is

used > > for > > > creating a translation, but the language model is used for

verifying

the > translation is good enough. If it does not validate against the

language > > model it should simply not be published to the main name space. It

will > > > still be possible to create a draft, but then the user is

completely

> > aware > > > that the translation isn't good enough. > > > > > > Such a language model should be available as a test for any

article,

as > > it > > > can be used as a quality measure for the article. It is really a > quantity > > > measure for the well-spokenness of the article, but that isn't

quite

> > intuitive. > > > > So, I'll allow myself to guess that you are talking about one

particular

> language, probably Norwegian. > > Several technical facts: > > 1. In the past there were several cases in which translators to

different > > languages who reported common translation mistakes to me. I passed

them

> on > > to Yandex developers, with whom I communicate quite regularly. They > > acknowledged receiving all of them. I am aware of at least one such > common > > mistake that was fixed; possibly there were more. If you can give me

list

machine > > translation from English is not available in Apertium. Apertium works > best > > with very similar languages, and English has two characteristics,

which

> are > > unfortunate when combined: it is both the most popular source for > > translation into almost all other languages (including Norwegian),

and

it > > is not _very_ similar to any other languages (except maybe Scots). > Machine > > translation from English into Norwegian is only possible with Yandex

the > moment. More engines may be added in the future, but at the moment

that's > > all we have. That's why disabling Yandex completely would indeed be a > lousy > > solution: A lot of people say that without machine translation > integration > > Content Translation is useless. Not all users think like that, but

many

> do. > > > > 3. We can define a numerical threshold of acceptable percentage of > machine > > translation post-editing. Currently it's 75%. It's a tad

embarrassing,

> but > > it's hard-coded at the moment, but it can be very easily be made

into a

> variable per language. If the translator tries to publish a page in

which > > less than that is modified, a warning will be shown. > > > > 4. I'm not sure what do you mean by "language model". If it's any

kind

of a > linguistic engine, then it's definitely not within the resources that

the > > Language team itself can currently dedicate. However, if somebody who > knows > > Norwegian and some programming will write a script that analyzes

common

> bad > > constructs in a Wikipedia dump, this will be very useful. This would > > basically be an upgraded version of suggestion #1 above. (In my spare > time > > as a volunteer I'm doing something comparable for Hebrew, although

not

> for > > translation, but for improving how MediaWiki link trails work.) > > _______________________________________________ > > Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ > > wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ > > wiki/Wikimedia-l > > New messages to: Wikimedia-l(a)lists.wikimedia.org > > Unsubscribe: https://lists.wikimedia.org/

mailman/listinfo/wikimedia-l,

<mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

Wojciech Pędzich

8:31 a.m.

It does depend a lot on the engagement level of the human behind the keyboard. When I deal with machine-translated text, I simply wonder whether the someone behind the keyboard took efforts to actually read the piece. Now whether this would work if limited to namespaces outside "main" - I do not want to demonise the issue, but if the person submitting the text for machine translation does not read it, what will stop them from a quick ctrl+c / ctrl+v? Just asking. Wojciech W dniu 2017-05-03 o 09:33, Yaroslav Blanter pisze:

...

This

Actually this _is_ about turning ContentTranslation off, that is what several users in the community want. They block people using the

extension > and delete the translated articles. Use of ContentTranslation has

become

rather contentious case. Yandex as a general translation engine to be able to read some alien language is quite good, but as an engine to produce written text it is

not > very good at all. In fact it often creates quite horrible Norwegian,

even

> for closely related languages. One quite common problem is reordering

> words into meaningless constructs, an other problem is reordering

lexical

> gender in weird ways. The English preposition "a" is often translated

> "en" in a propositional phrase, and then the gender is added to the > following phrase. That gives a translation of "Oppland is a county

in…"

> into something like "Oppland er en fylket i…" This should be "Oppland

use > Yandex and gets disappointed, and falsely conclude that all language > translations are equally weird. They are not, but Yandex translations

are

weird. The numerical threshold does not work. The reason is simple, the number

of > fixes depends on language constructs that fails, and that is simply

not a

> constant for small text fragments. Perhaps if we could flag specific > language constructs that is known to give a high percentage of

failures,

> and if the translator must check those sentences. One such language > construct is disappearances between the preposition and the gender of

the

following term in a prepositional phrase. If they are not similar, then

the

sentence must be checked. It is not always wrong to write "en jenta" in Norwegian, but it is likely to be wrong. A language model could be a statistical model for the language itself,

not > for the translation into that language. We don't want a perfect

language

> model, but a sufficient language model to mark weird constructs. A very > simple solution could simply be to mark tri-grams that does not

already

> exist in the text base for the destination as possible errors. It is

not

> necessary to do a live check, but at least do it before the page can

> saved. > > Note the difference in what Yandex do and what we want to achieve;

Yandex

translates a text between two different languages, without any clear

reason > why. It is not to important if there are weird constructs in the text,

long as it is usable in "some" context. We translate a text for the

purpose > of republishing it. The text should be usable and easily readable in

that

> language. > > > > On Tue, May 2, 2017 at 7:07 PM, Amir E. Aharoni < > amir.aharoni(a)mail.huji.ac.il> wrote: > >> 2017-05-02 18:20 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>: >> >>> Brute force solution; turn the ContentTranslation off. Really

stupid

>> solution. > > ... Then I guess you don't mind that I'm changing the thread name :) > > >> The next solution; turn the Yandex engine off. That would solve a >> part of the problem. Kind of lousy solution though. >> >> What about adding a language model that warns when the language > constructs >> gets to weird? It is like a "test" for the translation. The CT is

used >> for >>> creating a translation, but the language model is used for

verifying

> the >> translation is good enough. If it does not validate against the language >> model it should simply not be published to the main name space. It

will >>> still be possible to create a draft, but then the user is

completely

>> aware >>> that the translation isn't good enough. >>> >>> Such a language model should be available as a test for any

article,

as >> it >>> can be used as a quality measure for the article. It is really a > quantity >>> measure for the well-spokenness of the article, but that isn't

quite

>> intuitive. >> > So, I'll allow myself to guess that you are talking about one

particular

> language, probably Norwegian. > > Several technical facts: > > 1. In the past there were several cases in which translators to

different >> languages who reported common translation mistakes to me. I passed

them

> on >> to Yandex developers, with whom I communicate quite regularly. They >> acknowledged receiving all of them. I am aware of at least one such > common >> mistake that was fixed; possibly there were more. If you can give me

list > of such mistakes for Norwegian, I'll be very happy to pass them on. I > absolutely cannot promise that they will be fixed upstream, but it's > possible. > > 2. In Norwegian, Apertium is used for translating between the two varieties > of Norwegian itself (Bokmål and Nynorsk), and from other Scandinavian > languages. That's probably why it works so well—they are similar in > grammar, vocabulary, and narrative style (I'll pass it on to Apertium > developers—I'm sure they'll be happy to hear it). Unfortunately,

machine >> translation from English is not available in Apertium. Apertium works > best >> with very similar languages, and English has two characteristics,

which

> are >> unfortunate when combined: it is both the most popular source for >> translation into almost all other languages (including Norwegian),

and

it >> is not _very_ similar to any other languages (except maybe Scots). > Machine >> translation from English into Norwegian is only possible with Yandex

the > moment. More engines may be added in the future, but at the moment

that's >> all we have. That's why disabling Yandex completely would indeed be a > lousy >> solution: A lot of people say that without machine translation > integration >> Content Translation is useless. Not all users think like that, but

many

> do. >> 3. We can define a numerical threshold of acceptable percentage of > machine >> translation post-editing. Currently it's 75%. It's a tad

embarrassing,

> but >> it's hard-coded at the moment, but it can be very easily be made

into a

> variable per language. If the translator tries to publish a page in

which >> less than that is modified, a warning will be shown. >> >> 4. I'm not sure what do you mean by "language model". If it's any

kind

of a > linguistic engine, then it's definitely not within the resources that

the >> Language team itself can currently dedicate. However, if somebody who > knows >> Norwegian and some programming will write a script that analyzes

common

> bad >> constructs in a Wikipedia dump, this will be very useful. This would >> basically be an upgraded version of suggestion #1 above. (In my spare > time >> as a volunteer I'm doing something comparable for Hebrew, although

not

> for >> translation, but for improving how MediaWiki link trails work.) >> _______________________________________________ >> Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ >> wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ >> wiki/Wikimedia-l >> New messages to: Wikimedia-l(a)lists.wikimedia.org >> Unsubscribe: https://lists.wikimedia.org/

mailman/listinfo/wikimedia-l,

> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> _______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ wiki/Wikimedia-l New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

_______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/wiki/Wikimedia-l New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

David Cuenca Tudela

11:06 a.m.

Perhaps it would be a good idea to compare the translated text to the text that the user wants to save. If they are more than 95% the same, that means that the user didn't take the effort to correct the text. Cheers, Micru On Wed, May 3, 2017 at 10:31 AM, Wojciech Pędzich <wpedzich(a)gmail.com> wrote:

...

in the projects which do not have draft) could help. Cheers Yaroslav On Tue, May 2, 2017 at 10:16 PM, Pharos <pharosofalexandria(a)gmail.com> wrote: I think it all depends on the level of engagement of the human translator.

When the tool is used in the right way, it is a fantastic tool. Maybe we can find better methods to nudge people toward taking their time and really doing work on their translations. Thanks, Pharos On Tue, May 2, 2017 at 4:09 PM, Bodhisattwa Mandal < bodhisattwa.rgkmc(a)gmail.com> wrote: Content translation with Yandex is also a problem in Bengali Wikipedia.

Some users have grown a tendency to create machine translated meaningless articles with this extension to increase edit count and article count.

This

has increased the workloads of admins to find and delete those articles. Yandex is not ready for many languages and it is better to shut it. We don't need it in Bengali. Regards On May 3, 2017 12:17 AM, "John Erling Blad" <jeblad(a)gmail.com> wrote: Actually this _is_ about turning ContentTranslation off, that is what > several users in the community want. They block people using the > extension > and delete the translated articles. Use of ContentTranslation has > become

a > rather contentious case. > > Yandex as a general translation engine to be able to read some alien > language is quite good, but as an engine to produce written text it is > not > very good at all. In fact it often creates quite horrible Norwegian, > even

for closely related languages. One quite common problem is reordering > of

words into meaningless constructs, an other problem is reordering > lexical

gender in weird ways. The English preposition "a" is often translated > as

"en" in a propositional phrase, and then the gender is added to the > following phrase. That gives a translation of "Oppland is a county > in…"

into something like "Oppland er en fylket i…" This should be "Oppland > er

et fylke i…". > > (I just checked and it seems like Yandex messes up a lot less now than > previously, but it is still pretty bad.) > > Apertium works because the language is closely related, Yandex does not > work because it is used between very different languages. People try to > use > Yandex and gets disappointed, and falsely conclude that all language > translations are equally weird. They are not, but Yandex translations > are

weird. > > The numerical threshold does not work. The reason is simple, the number > of > fixes depends on language constructs that fails, and that is simply > not a

constant for small text fragments. Perhaps if we could flag specific > language constructs that is known to give a high percentage of > failures,

and if the translator must check those sentences. One such language > construct is disappearances between the preposition and the gender of > the

following term in a prepositional phrase. If they are not similar, then > the > sentence must be checked. It is not always wrong to write "en jenta" in > Norwegian, but it is likely to be wrong. > > A language model could be a statistical model for the language itself, > not > for the translation into that language. We don't want a perfect > language

model, but a sufficient language model to mark weird constructs. A very > simple solution could simply be to mark tri-grams that does not > already

exist in the text base for the destination as possible errors. It is > not

necessary to do a live check, but at least do it before the page can > be

saved. > > Note the difference in what Yandex do and what we want to achieve; > Yandex

translates a text between two different languages, without any clear > reason > why. It is not to important if there are weird constructs in the text, > as

long as it is usable in "some" context. We translate a text for the > purpose > of republishing it. The text should be usable and easily readable in > that

language. > > > > On Tue, May 2, 2017 at 7:07 PM, Amir E. Aharoni < > amir.aharoni(a)mail.huji.ac.il> wrote: > > 2017-05-02 18:20 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>: >> >> Brute force solution; turn the ContentTranslation off. Really >>> >> stupid

solution. >>> >> >> ... Then I guess you don't mind that I'm changing the thread name :) >> >> >> The next solution; turn the Yandex engine off. That would solve a >>> part of the problem. Kind of lousy solution though. >>> >>> What about adding a language model that warns when the language >>> >> constructs >> >>> gets to weird? It is like a "test" for the translation. The CT is >>> >> used > for >> >>> creating a translation, but the language model is used for >>> >> verifying

if > the >> >>> translation is good enough. If it does not validate against the >>> >> language > >> model it should simply not be published to the main name space. It >>> >> will > still be possible to create a draft, but then the user is >>> >> completely

aware >> >>> that the translation isn't good enough. >>> >>> Such a language model should be available as a test for any >>> >> article,

as > it >> >>> can be used as a quality measure for the article. It is really a >>> >> quantity > >> measure for the well-spokenness of the article, but that isn't >>> >> quite

so > intuitive. >>> >>> So, I'll allow myself to guess that you are talking about one >> > particular > language, probably Norwegian. >> >> Several technical facts: >> >> 1. In the past there were several cases in which translators to >> > different > languages who reported common translation mistakes to me. I passed >> > them

on > >> to Yandex developers, with whom I communicate quite regularly. They >> acknowledged receiving all of them. I am aware of at least one such >> > common > >> mistake that was fixed; possibly there were more. If you can give me >> > a

list > >> of such mistakes for Norwegian, I'll be very happy to pass them on. I >> absolutely cannot promise that they will be fixed upstream, but it's >> possible. >> >> 2. In Norwegian, Apertium is used for translating between the two >> > varieties > >> of Norwegian itself (Bokmål and Nynorsk), and from other Scandinavian >> languages. That's probably why it works so well—they are similar in >> grammar, vocabulary, and narrative style (I'll pass it on to Apertium >> developers—I'm sure they'll be happy to hear it). Unfortunately, >> > machine > translation from English is not available in Apertium. Apertium works >> > best > >> with very similar languages, and English has two characteristics, >> > which

are > >> unfortunate when combined: it is both the most popular source for >> translation into almost all other languages (including Norwegian), >> > and

it > is not _very_ similar to any other languages (except maybe Scots). >> > Machine > >> translation from English into Norwegian is only possible with Yandex >> > at

the > >> moment. More engines may be added in the future, but at the moment >> > that's > all we have. That's why disabling Yandex completely would indeed be a >> > lousy > >> solution: A lot of people say that without machine translation >> > integration > >> Content Translation is useless. Not all users think like that, but >> > many

do. > >> 3. We can define a numerical threshold of acceptable percentage of >> > machine > >> translation post-editing. Currently it's 75%. It's a tad >> > embarrassing,

but > >> it's hard-coded at the moment, but it can be very easily be made >> > into a

variable per language. If the translator tries to publish a page in >> > which > less than that is modified, a warning will be shown. >> >> 4. I'm not sure what do you mean by "language model". If it's any >> > kind

of a > >> linguistic engine, then it's definitely not within the resources that >> > the > Language team itself can currently dedicate. However, if somebody who >> > knows > >> Norwegian and some programming will write a script that analyzes >> > common

bad > >> constructs in a Wikipedia dump, this will be very useful. This would >> basically be an upgraded version of suggestion #1 above. (In my spare >> > time > >> as a volunteer I'm doing something comparable for Hebrew, although >> > not

for > >> translation, but for improving how MediaWiki link trails work.) >> _______________________________________________ >> Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ >> wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ >> wiki/Wikimedia-l >> New messages to: Wikimedia-l(a)lists.wikimedia.org >> Unsubscribe: https://lists.wikimedia.org/ >> > mailman/listinfo/wikimedia-l,

<mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> >> > _______________________________________________ > Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ > wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ > wiki/Wikimedia-l > New messages to: Wikimedia-l(a)lists.wikimedia.org > Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, > <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> > _______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ wiki/Wikimedia-l New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> _______________________________________________

Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ wiki/Wikimedia-l New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> _______________________________________________

Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wik i/Mailing_lists/Guidelines and https://meta.wikimedia.org/wik i/Wikimedia-l New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

_______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wik i/Mailing_lists/Guidelines and https://meta.wikimedia.org/wiki/Wikimedia-l New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

-- Etiamsi omnes, ego non

Lodewijk

11:15 a.m.

Reading this, I get a strong impression the problem may very well be in setting expectations for the users of this translation tool. If they expect the automated translation to be rather good, they may get fed up more easily than when they consider it primarily a glorified dictionary. Lodewijk On Wed, May 3, 2017 at 1:06 PM, David Cuenca Tudela <dacuetu(a)gmail.com> wrote:

...

It does depend a lot on the engagement level of the human behind the keyboard. When I deal with machine-translated text, I simply wonder

whether

the someone behind the keyboard took efforts to actually read the piece. Now whether this would work if limited to namespaces outside "main" - I

not want to demonise the issue, but if the person submitting the text for machine translation does not read it, what will stop them from a quick ctrl+c / ctrl+v? Just asking. Wojciech W dniu 2017-05-03 o 09:33, Yaroslav Blanter pisze: Creating machine translations only in the draft space (or in the user

space

> in the projects which do not have draft) could help. > > Cheers > Yaroslav > > On Tue, May 2, 2017 at 10:16 PM, Pharos <pharosofalexandria(a)gmail.com> > wrote: > > I think it all depends on the level of engagement of the human

translator.

>> >> When the tool is used in the right way, it is a fantastic tool. >> >> Maybe we can find better methods to nudge people toward taking their

time

>> and really doing work on their translations. >> >> Thanks, >> Pharos >> >> On Tue, May 2, 2017 at 4:09 PM, Bodhisattwa Mandal < >> bodhisattwa.rgkmc(a)gmail.com> wrote: >> >> Content translation with Yandex is also a problem in Bengali Wikipedia. >>> Some users have grown a tendency to create machine translated >>> meaningless >>> articles with this extension to increase edit count and article count. >>> >> This >> >>> has increased the workloads of admins to find and delete those

articles.

>>> >>> Yandex is not ready for many languages and it is better to shut it. We >>> don't need it in Bengali. >>> >>> Regards >>> On May 3, 2017 12:17 AM, "John Erling Blad" <jeblad(a)gmail.com> wrote: >>> >>> Actually this _is_ about turning ContentTranslation off, that is what >>>> several users in the community want. They block people using the >>>> >>> extension >>> >>>> and delete the translated articles. Use of ContentTranslation has >>>> >>> become >> >>> a >>> >>>> rather contentious case. >>>> >>>> Yandex as a general translation engine to be able to read some alien >>>> language is quite good, but as an engine to produce written text it

>>>> >>> not >>> >>>> very good at all. In fact it often creates quite horrible Norwegian, >>>> >>> even >> >>> for closely related languages. One quite common problem is reordering >>>> >>> of >> >>> words into meaningless constructs, an other problem is reordering >>>> >>> lexical >> >>> gender in weird ways. The English preposition "a" is often translated >>>> >>> as >> >>> "en" in a propositional phrase, and then the gender is added to the >>>> following phrase. That gives a translation of "Oppland is a county >>>> >>> in…" >> >>> into something like "Oppland er en fylket i…" This should be

"Oppland

>>>> >>> er >> >>> et fylke i…". >>>> >>>> (I just checked and it seems like Yandex messes up a lot less now

than

>>>> previously, but it is still pretty bad.) >>>> >>>> Apertium works because the language is closely related, Yandex does

not

>>>> work because it is used between very different languages. People try

>>>> >>> use >>> >>>> Yandex and gets disappointed, and falsely conclude that all language >>>> translations are equally weird. They are not, but Yandex translations >>>> >>> are >> >>> weird. >>>> >>>> The numerical threshold does not work. The reason is simple, the

number

>>>> >>> of >>> >>>> fixes depends on language constructs that fails, and that is simply >>>> >>> not a >> >>> constant for small text fragments. Perhaps if we could flag specific >>>> language constructs that is known to give a high percentage of >>>> >>> failures, >> >>> and if the translator must check those sentences. One such language >>>> construct is disappearances between the preposition and the gender of >>>> >>> the >> >>> following term in a prepositional phrase. If they are not similar,

then

>>>> >>> the >>> >>>> sentence must be checked. It is not always wrong to write "en jenta"

>>>> Norwegian, but it is likely to be wrong. >>>> >>>> A language model could be a statistical model for the language

itself,

>>>> >>> not >>> >>>> for the translation into that language. We don't want a perfect >>>> >>> language >> >>> model, but a sufficient language model to mark weird constructs. A

very

>>>> simple solution could simply be to mark tri-grams that does not >>>> >>> already >> >>> exist in the text base for the destination as possible errors. It is >>>> >>> not >> >>> necessary to do a live check, but at least do it before the page can >>>> >>> be >> >>> saved. >>>> >>>> Note the difference in what Yandex do and what we want to achieve; >>>> >>> Yandex >> >>> translates a text between two different languages, without any clear >>>> >>> reason >>> >>>> why. It is not to important if there are weird constructs in the

text,

>>>> >>> as >> >>> long as it is usable in "some" context. We translate a text for the >>>> >>> purpose >>> >>>> of republishing it. The text should be usable and easily readable in >>>> >>> that >> >>> language. >>>> >>>> >>>> >>>> On Tue, May 2, 2017 at 7:07 PM, Amir E. Aharoni < >>>> amir.aharoni(a)mail.huji.ac.il> wrote: >>>> >>>> 2017-05-02 18:20 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>: >>>>> >>>>> Brute force solution; turn the ContentTranslation off. Really >>>>>> >>>>> stupid >> >>> solution. >>>>>> >>>>> >>>>> ... Then I guess you don't mind that I'm changing the thread name :) >>>>> >>>>> >>>>> The next solution; turn the Yandex engine off. That would solve a >>>>>> part of the problem. Kind of lousy solution though. >>>>>> >>>>>> What about adding a language model that warns when the language >>>>>> >>>>> constructs >>>>> >>>>>> gets to weird? It is like a "test" for the translation. The CT is >>>>>> >>>>> used >>> >>>> for >>>>> >>>>>> creating a translation, but the language model is used for >>>>>> >>>>> verifying >> >>> if >>> >>>> the >>>>> >>>>>> translation is good enough. If it does not validate against the >>>>>> >>>>> language >>>> >>>>> model it should simply not be published to the main name space. It >>>>>> >>>>> will >>> >>>> still be possible to create a draft, but then the user is >>>>>> >>>>> completely >> >>> aware >>>>> >>>>>> that the translation isn't good enough. >>>>>> >>>>>> Such a language model should be available as a test for any >>>>>> >>>>> article, >> >>> as >>> >>>> it >>>>> >>>>>> can be used as a quality measure for the article. It is really a >>>>>> >>>>> quantity >>>> >>>>> measure for the well-spokenness of the article, but that isn't >>>>>> >>>>> quite >> >>> so >>> >>>> intuitive. >>>>>> >>>>>> So, I'll allow myself to guess that you are talking about one >>>>> >>>> particular >>> >>>> language, probably Norwegian. >>>>> >>>>> Several technical facts: >>>>> >>>>> 1. In the past there were several cases in which translators to >>>>> >>>> different >>> >>>> languages who reported common translation mistakes to me. I passed >>>>> >>>> them >> >>> on >>>> >>>>> to Yandex developers, with whom I communicate quite regularly. They >>>>> acknowledged receiving all of them. I am aware of at least one such >>>>> >>>> common >>>> >>>>> mistake that was fixed; possibly there were more. If you can give me >>>>> >>>> a >> >>> list >>>> >>>>> of such mistakes for Norwegian, I'll be very happy to pass them on.

>>>>> absolutely cannot promise that they will be fixed upstream, but it's >>>>> possible. >>>>> >>>>> 2. In Norwegian, Apertium is used for translating between the two >>>>> >>>> varieties >>>> >>>>> of Norwegian itself (Bokmål and Nynorsk), and from other

Scandinavian

>>>>> languages. That's probably why it works so well—they are similar in >>>>> grammar, vocabulary, and narrative style (I'll pass it on to

Apertium

>>>>> developers—I'm sure they'll be happy to hear it). Unfortunately, >>>>> >>>> machine >>> >>>> translation from English is not available in Apertium. Apertium works >>>>> >>>> best >>>> >>>>> with very similar languages, and English has two characteristics, >>>>> >>>> which >> >>> are >>>> >>>>> unfortunate when combined: it is both the most popular source for >>>>> translation into almost all other languages (including Norwegian), >>>>> >>>> and >> >>> it >>> >>>> is not _very_ similar to any other languages (except maybe Scots). >>>>> >>>> Machine >>>> >>>>> translation from English into Norwegian is only possible with Yandex >>>>> >>>> at >> >>> the >>>> >>>>> moment. More engines may be added in the future, but at the moment >>>>> >>>> that's >>> >>>> all we have. That's why disabling Yandex completely would indeed be a >>>>> >>>> lousy >>>> >>>>> solution: A lot of people say that without machine translation >>>>> >>>> integration >>>> >>>>> Content Translation is useless. Not all users think like that, but >>>>> >>>> many >> >>> do. >>>> >>>>> 3. We can define a numerical threshold of acceptable percentage of >>>>> >>>> machine >>>> >>>>> translation post-editing. Currently it's 75%. It's a tad >>>>> >>>> embarrassing, >> >>> but >>>> >>>>> it's hard-coded at the moment, but it can be very easily be made >>>>> >>>> into a >> >>> variable per language. If the translator tries to publish a page in >>>>> >>>> which >>> >>>> less than that is modified, a warning will be shown. >>>>> >>>>> 4. I'm not sure what do you mean by "language model". If it's any >>>>> >>>> kind >> >>> of a >>>> >>>>> linguistic engine, then it's definitely not within the resources

that

>>>>> >>>> the >>> >>>> Language team itself can currently dedicate. However, if somebody who >>>>> >>>> knows >>>> >>>>> Norwegian and some programming will write a script that analyzes >>>>> >>>> common >> >>> bad >>>> >>>>> constructs in a Wikipedia dump, this will be very useful. This would >>>>> basically be an upgraded version of suggestion #1 above. (In my

spare

>>>>> >>>> time >>>> >>>>> as a volunteer I'm doing something comparable for Hebrew, although >>>>> >>>> not >> >>> for >>>> >>>>> translation, but for improving how MediaWiki link trails work.) >>>>> _______________________________________________ >>>>> Wikimedia-l mailing list, guidelines at:

https://meta.wikimedia.org/

>>>>> wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ >>>>> wiki/Wikimedia-l >>>>> New messages to: Wikimedia-l(a)lists.wikimedia.org >>>>> Unsubscribe: https://lists.wikimedia.org/ >>>>> >>>> mailman/listinfo/wikimedia-l, >> >>> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> >>>>> >>>> _______________________________________________ >>>> Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ >>>> wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ >>>> wiki/Wikimedia-l >>>> New messages to: Wikimedia-l(a)lists.wikimedia.org >>>> Unsubscribe: https://lists.wikimedia.org/

mailman/listinfo/wikimedia-l,

>>>> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> >>>> >>> _______________________________________________ >>> Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ >>> wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ >>> wiki/Wikimedia-l >>> New messages to: Wikimedia-l(a)lists.wikimedia.org >>> Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l

> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> > > _______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ wiki/Wikimedia-l New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> _______________________________________________

Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wik i/Mailing_lists/Guidelines and https://meta.wikimedia.org/wik i/Wikimedia-l New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

_______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wik i/Mailing_lists/Guidelines and https://meta.wikimedia.org/

wiki/Wikimedia-l

New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

-- Etiamsi omnes, ego non _______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ wiki/Wikimedia-l New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

John Erling Blad

11:22 a.m.

Agree! I also wonder if translators adapt to specific errors if they are repeated to often. I wonder if it works like priming the brain to a specific pattern. On Wed, May 3, 2017 at 1:15 PM, Lodewijk <lodewijk(a)effeietsanders.org> wrote:

...

Perhaps it would be a good idea to compare the translated text to the

text

that the user wants to save. If they are more than 95% the same, that means that the user didn't take the effort to correct the text. Cheers, Micru On Wed, May 3, 2017 at 10:31 AM, Wojciech Pędzich <wpedzich(a)gmail.com> wrote:

It does depend a lot on the engagement level of the human behind the keyboard. When I deal with machine-translated text, I simply wonder

whether > the someone behind the keyboard took efforts to actually read the

piece.

Now whether this would work if limited to namespaces outside "main" - I

do > not want to demonise the issue, but if the person submitting the text

for

machine translation does not read it, what will stop them from a quick ctrl+c / ctrl+v? Just asking. Wojciech W dniu 2017-05-03 o 09:33, Yaroslav Blanter pisze: Creating machine translations only in the draft space (or in the user

space

> in the projects which do not have draft) could help. > > Cheers > Yaroslav > > On Tue, May 2, 2017 at 10:16 PM, Pharos <pharosofalexandria(a)gmail.com

> wrote: > > I think it all depends on the level of engagement of the human

translator.

>> >> When the tool is used in the right way, it is a fantastic tool. >> >> Maybe we can find better methods to nudge people toward taking their

time >>> and really doing work on their translations. >>> >>> Thanks, >>> Pharos >>> >>> On Tue, May 2, 2017 at 4:09 PM, Bodhisattwa Mandal < >>> bodhisattwa.rgkmc(a)gmail.com> wrote: >>> >>> Content translation with Yandex is also a problem in Bengali

Wikipedia.

>>>> Some users have grown a tendency to create machine translated >>>> meaningless >>>> articles with this extension to increase edit count and article

count.

>>> >> This >> >>> has increased the workloads of admins to find and delete those

articles. >>>> >>>> Yandex is not ready for many languages and it is better to shut it.

>>>> don't need it in Bengali. >>>> >>>> Regards >>>> On May 3, 2017 12:17 AM, "John Erling Blad" <jeblad(a)gmail.com>

wrote:

>>>> >>>> Actually this _is_ about turning ContentTranslation off, that is

what

>>>>> several users in the community want. They block people using the >>>>> >>>> extension >>>> >>>>> and delete the translated articles. Use of ContentTranslation has >>>>> >>>> become >>> >>>> a >>>> >>>>> rather contentious case. >>>>> >>>>> Yandex as a general translation engine to be able to read some

alien

>>>> language is quite good, but as an engine to produce written text it

is >>>>> >>>> not >>>> >>>>> very good at all. In fact it often creates quite horrible

Norwegian,

>>>>> >>>> even >>> >>>> for closely related languages. One quite common problem is

reordering

>>>>> >>>> of >>> >>>> words into meaningless constructs, an other problem is reordering >>>>> >>>> lexical >>> >>>> gender in weird ways. The English preposition "a" is often

translated

>>>> >>> as >> >>> "en" in a propositional phrase, and then the gender is added to the >>>> following phrase. That gives a translation of "Oppland is a county >>>> >>> in…" >> >>> into something like "Oppland er en fylket i…" This should be

"Oppland

>>>> >>> er >> >>> et fylke i…". >>>> >>>> (I just checked and it seems like Yandex messes up a lot less now

than

>>>> previously, but it is still pretty bad.) >>>> >>>> Apertium works because the language is closely related, Yandex does

not >>>>> work because it is used between very different languages. People

try

to >>>>> >>>> use >>>> >>>>> Yandex and gets disappointed, and falsely conclude that all

language

>>>>> translations are equally weird. They are not, but Yandex

translations

>>>> >>> are >> >>> weird. >>>> >>>> The numerical threshold does not work. The reason is simple, the

number >>>>> >>>> of >>>> >>>>> fixes depends on language constructs that fails, and that is simply >>>>> >>>> not a >>> >>>> constant for small text fragments. Perhaps if we could flag specific >>>>> language constructs that is known to give a high percentage of >>>>> >>>> failures, >>> >>>> and if the translator must check those sentences. One such language >>>>> construct is disappearances between the preposition and the gender

>>>> >>> the >> >>> following term in a prepositional phrase. If they are not similar,

then >>>>> >>>> the >>>> >>>>> sentence must be checked. It is not always wrong to write "en

jenta"

>>>> Norwegian, but it is likely to be wrong. >>>> >>>> A language model could be a statistical model for the language

itself,

>>>> >>> not >>> >>>> for the translation into that language. We don't want a perfect >>>> >>> language >> >>> model, but a sufficient language model to mark weird constructs. A

very >>>>> simple solution could simply be to mark tri-grams that does not >>>>> >>>> already >>> >>>> exist in the text base for the destination as possible errors. It is >>>>> >>>> not >>> >>>> necessary to do a live check, but at least do it before the page

can

>>>> >>> be >> >>> saved. >>>> >>>> Note the difference in what Yandex do and what we want to achieve; >>>> >>> Yandex >> >>> translates a text between two different languages, without any clear >>>> >>> reason >>> >>>> why. It is not to important if there are weird constructs in the

text, >>>>> >>>> as >>> >>>> long as it is usable in "some" context. We translate a text for the >>>>> >>>> purpose >>>> >>>>> of republishing it. The text should be usable and easily readable

>>>>> >>>> that >>> >>>> language. >>>>> >>>>> >>>>> >>>>> On Tue, May 2, 2017 at 7:07 PM, Amir E. Aharoni < >>>>> amir.aharoni(a)mail.huji.ac.il> wrote: >>>>> >>>>> 2017-05-02 18:20 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>: >>>>>> >>>>>> Brute force solution; turn the ContentTranslation off. Really >>>>>>> >>>>>> stupid >>> >>>> solution. >>>>>>> >>>>>> >>>>>> ... Then I guess you don't mind that I'm changing the thread name

>>>>>> >>>>>> >>>>>> The next solution; turn the Yandex engine off. That would solve a >>>>>>> part of the problem. Kind of lousy solution though. >>>>>>> >>>>>>> What about adding a language model that warns when the language >>>>>>> >>>>>> constructs >>>>>> >>>>>>> gets to weird? It is like a "test" for the translation. The CT is >>>>>>> >>>>>> used >>>> >>>>> for >>>>>> >>>>>>> creating a translation, but the language model is used for >>>>>>> >>>>>> verifying >>> >>>> if >>>> >>>>> the >>>>>> >>>>>>> translation is good enough. If it does not validate against the >>>>>>> >>>>>> language >>>>> >>>>>> model it should simply not be published to the main name space. It >>>>>>> >>>>>> will >>>> >>>>> still be possible to create a draft, but then the user is >>>>>>> >>>>>> completely >>> >>>> aware >>>>>> >>>>>>> that the translation isn't good enough. >>>>>>> >>>>>>> Such a language model should be available as a test for any >>>>>>> >>>>>> article, >>> >>>> as >>>> >>>>> it >>>>>> >>>>>>> can be used as a quality measure for the article. It is really a >>>>>>> >>>>>> quantity >>>>> >>>>>> measure for the well-spokenness of the article, but that isn't >>>>>>> >>>>>> quite >>> >>>> so >>>> >>>>> intuitive. >>>>>>> >>>>>>> So, I'll allow myself to guess that you are talking about one >>>>>> >>>>> particular >>>> >>>>> language, probably Norwegian. >>>>>> >>>>>> Several technical facts: >>>>>> >>>>>> 1. In the past there were several cases in which translators to >>>>>> >>>>> different >>>> >>>>> languages who reported common translation mistakes to me. I passed >>>>>> >>>>> them >>> >>>> on >>>>> >>>>>> to Yandex developers, with whom I communicate quite regularly.

They

>>>>>> acknowledged receiving all of them. I am aware of at least one

such

>>>>>> >>>>> common >>>>> >>>>>> mistake that was fixed; possibly there were more. If you can give

>>>>>> >>>>> a >>> >>>> list >>>>> >>>>>> of such mistakes for Norwegian, I'll be very happy to pass them

on.

I >>>>>> absolutely cannot promise that they will be fixed upstream, but

it's

>>>>> possible. >>>>> >>>>> 2. In Norwegian, Apertium is used for translating between the two >>>>> >>>> varieties >>>> >>>>> of Norwegian itself (Bokmål and Nynorsk), and from other

Scandinavian >>>>>> languages. That's probably why it works so well—they are similar

>>>>> grammar, vocabulary, and narrative style (I'll pass it on to

Apertium >>>>>> developers—I'm sure they'll be happy to hear it). Unfortunately, >>>>>> >>>>> machine >>>> >>>>> translation from English is not available in Apertium. Apertium

works

>>>>>> >>>>> best >>>>> >>>>>> with very similar languages, and English has two characteristics, >>>>>> >>>>> which >>> >>>> are >>>>> >>>>>> unfortunate when combined: it is both the most popular source for >>>>>> translation into almost all other languages (including Norwegian), >>>>>> >>>>> and >>> >>>> it >>>> >>>>> is not _very_ similar to any other languages (except maybe Scots). >>>>>> >>>>> Machine >>>>> >>>>>> translation from English into Norwegian is only possible with

Yandex

>>>>>> >>>>> at >>> >>>> the >>>>> >>>>>> moment. More engines may be added in the future, but at the moment >>>>>> >>>>> that's >>>> >>>>> all we have. That's why disabling Yandex completely would indeed

be a

>>>>> >>>> lousy >>>> >>>>> solution: A lot of people say that without machine translation >>>>> >>>> integration >>>> >>>>> Content Translation is useless. Not all users think like that, but >>>>> >>>> many >> >>> do. >>>> >>>>> 3. We can define a numerical threshold of acceptable percentage of >>>>> >>>> machine >>>> >>>>> translation post-editing. Currently it's 75%. It's a tad >>>>> >>>> embarrassing, >> >>> but >>>> >>>>> it's hard-coded at the moment, but it can be very easily be made >>>>> >>>> into a >> >>> variable per language. If the translator tries to publish a page in >>>>> >>>> which >>> >>>> less than that is modified, a warning will be shown. >>>>> >>>>> 4. I'm not sure what do you mean by "language model". If it's any >>>>> >>>> kind >> >>> of a >>>> >>>>> linguistic engine, then it's definitely not within the resources

that >>>>>> >>>>> the >>>> >>>>> Language team itself can currently dedicate. However, if somebody

who

>>>>>> >>>>> knows >>>>> >>>>>> Norwegian and some programming will write a script that analyzes >>>>>> >>>>> common >>> >>>> bad >>>>> >>>>>> constructs in a Wikipedia dump, this will be very useful. This

would

>>>>> basically be an upgraded version of suggestion #1 above. (In my

spare

https://meta.wikimedia.org/ >>>>>> wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ >>>>>> wiki/Wikimedia-l >>>>>> New messages to: Wikimedia-l(a)lists.wikimedia.org >>>>>> Unsubscribe: https://lists.wikimedia.org/ >>>>>> >>>>> mailman/listinfo/wikimedia-l, >>> >>>> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=

unsubscribe>

>>>>>> >>>>> _______________________________________________ >>>>> Wikimedia-l mailing list, guidelines at:

https://meta.wikimedia.org/

>>>> wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ >>>> wiki/Wikimedia-l >>>> New messages to: Wikimedia-l(a)lists.wikimedia.org >>>> Unsubscribe: https://lists.wikimedia.org/

mailman/listinfo/wikimedia-l, >>>>> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=

unsubscribe>

>>>>> >>>> _______________________________________________ >>>> Wikimedia-l mailing list, guidelines at:

https://meta.wikimedia.org/

>>>> wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ >>>> wiki/Wikimedia-l >>>> New messages to: Wikimedia-l(a)lists.wikimedia.org >>>> Unsubscribe: https://lists.wikimedia.org/

mailman/listinfo/wikimedia-l

, >>>> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=

unsubscribe>

>>>> >>>> _______________________________________________ >>> Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/ >>> wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ >>> wiki/Wikimedia-l >>> New messages to: Wikimedia-l(a)lists.wikimedia.org >>> Unsubscribe: https://lists.wikimedia.org/

mailman/listinfo/wikimedia-l,

>>> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> >>> >>> _______________________________________________ >> Wikimedia-l mailing list, guidelines at:

https://meta.wikimedia.org/wik

>> i/Mailing_lists/Guidelines and https://meta.wikimedia.org/wik >> i/Wikimedia-l >> New messages to: Wikimedia-l(a)lists.wikimedia.org >> Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l

>> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> >> > > > > _______________________________________________ > Wikimedia-l mailing list, guidelines at:

https://meta.wikimedia.org/wik

i/Mailing_lists/Guidelines and https://meta.wikimedia.org/

wiki/Wikimedia-l

New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

Ziko van Dijk

11:27 a.m.

Hello, This seems to me like a social problem, rather than a technical one. Shutting down the tool would be a disadvantage for those people who benefit from the tool and do good things with it. What is the general opinion among the Norwegians about this issue? Is there consent about how to deal with this kind of "articles"? If most people agree they should be speedy-deleted, this would be a useful deterrence for those who are not careful enough when using the tool? Kind regards Ziko 2017-05-03 13:22 GMT+02:00 John Erling Blad <jeblad(a)gmail.com>om>:

...

Reading this, I get a strong impression the problem may very well be in setting expectations for the users of this translation tool. If they

expect

the automated translation to be rather good, they may get fed up more easily than when they consider it primarily a glorified dictionary. Lodewijk On Wed, May 3, 2017 at 1:06 PM, David Cuenca Tudela <dacuetu(a)gmail.com> wrote:

Perhaps it would be a good idea to compare the translated text to the

text > that the user wants to save. > > If they are more than 95% the same, that means that the user didn't

take

the effort to correct the text. Cheers, Micru On Wed, May 3, 2017 at 10:31 AM, Wojciech Pędzich <wpedzich(a)gmail.com> wrote:

It does depend a lot on the engagement level of the human behind the keyboard. When I deal with machine-translated text, I simply wonder

whether > the someone behind the keyboard took efforts to actually read the

piece. > > > > Now whether this would work if limited to namespaces outside "main"

- I

do > not want to demonise the issue, but if the person submitting the text

for > > machine translation does not read it, what will stop them from a

quick

> > ctrl+c / ctrl+v? Just asking. > > > > Wojciech > > > > W dniu 2017-05-03 o 09:33, Yaroslav Blanter pisze: > > > > Creating machine translations only in the draft space (or in the user > space > >> in the projects which do not have draft) could help. > >> > >> Cheers > >> Yaroslav > >> > >> On Tue, May 2, 2017 at 10:16 PM, Pharos <

pharosofalexandria(a)gmail.com

> > >> wrote: > >> > >> I think it all depends on the level of engagement of the human > translator. > >>> > >>> When the tool is used in the right way, it is a fantastic tool. > >>> > >>> Maybe we can find better methods to nudge people toward taking

their

Wikipedia.

>>>> Some users have grown a tendency to create machine translated >>>> meaningless >>>> articles with this extension to increase edit count and article

count. > >>>> > >>> This > >>> > >>>> has increased the workloads of admins to find and delete those > articles. > >>>> > >>>> Yandex is not ready for many languages and it is better to shut

it.

>>>> don't need it in Bengali. >>>> >>>> Regards >>>> On May 3, 2017 12:17 AM, "John Erling Blad" <jeblad(a)gmail.com>

wrote:

>>>> >>>> Actually this _is_ about turning ContentTranslation off, that is

what

alien > >>>>> language is quite good, but as an engine to produce written text

is >>>>> >>>> not >>>> >>>>> very good at all. In fact it often creates quite horrible

Norwegian,

>>>>> >>>> even >>> >>>> for closely related languages. One quite common problem is

reordering

>>>>> >>>> of >>> >>>> words into meaningless constructs, an other problem is reordering >>>>> >>>> lexical >>> >>>> gender in weird ways. The English preposition "a" is often

translated > >>>>> > >>>> as > >>> > >>>> "en" in a propositional phrase, and then the gender is added to

the

> >>>>> following phrase. That gives a translation of "Oppland is a

county

> >>>>> > >>>> in…" > >>> > >>>> into something like "Oppland er en fylket i…" This should be > "Oppland > >>>>> > >>>> er > >>> > >>>> et fylke i…". > >>>>> > >>>>> (I just checked and it seems like Yandex messes up a lot less now > than > >>>>> previously, but it is still pretty bad.) > >>>>> > >>>>> Apertium works because the language is closely related, Yandex

does

not >>>>> work because it is used between very different languages. People

try

to >>>>> >>>> use >>>> >>>>> Yandex and gets disappointed, and falsely conclude that all

language

>>>>> translations are equally weird. They are not, but Yandex

translations > >>>>> > >>>> are > >>> > >>>> weird. > >>>>> > >>>>> The numerical threshold does not work. The reason is simple, the > number > >>>>> > >>>> of > >>>> > >>>>> fixes depends on language constructs that fails, and that is

simply

> >>>>> > >>>> not a > >>> > >>>> constant for small text fragments. Perhaps if we could flag

specific

> >>>>> language constructs that is known to give a high percentage of > >>>>> > >>>> failures, > >>> > >>>> and if the translator must check those sentences. One such

language

> >>>>> construct is disappearances between the preposition and the

gender

>>>> >>> the >> >>> following term in a prepositional phrase. If they are not similar,

then >>>>> >>>> the >>>> >>>>> sentence must be checked. It is not always wrong to write "en

jenta" > in > >>>>> Norwegian, but it is likely to be wrong. > >>>>> > >>>>> A language model could be a statistical model for the language > itself, > >>>>> > >>>> not > >>>> > >>>>> for the translation into that language. We don't want a perfect > >>>>> > >>>> language > >>> > >>>> model, but a sufficient language model to mark weird constructs. A > very > >>>>> simple solution could simply be to mark tri-grams that does not > >>>>> > >>>> already > >>> > >>>> exist in the text base for the destination as possible errors. It

>>>>> >>>> not >>> >>>> necessary to do a live check, but at least do it before the page

can > >>>>> > >>>> be > >>> > >>>> saved. > >>>>> > >>>>> Note the difference in what Yandex do and what we want to

achieve;

> >>>>> > >>>> Yandex > >>> > >>>> translates a text between two different languages, without any

clear

> >>>>> > >>>> reason > >>>> > >>>>> why. It is not to important if there are weird constructs in the > text, > >>>>> > >>>> as > >>> > >>>> long as it is usable in "some" context. We translate a text for

the

>>>>> >>>> purpose >>>> >>>>> of republishing it. The text should be usable and easily readable

in > >>>>> > >>>> that > >>> > >>>> language. > >>>>> > >>>>> > >>>>> > >>>>> On Tue, May 2, 2017 at 7:07 PM, Amir E. Aharoni < > >>>>> amir.aharoni(a)mail.huji.ac.il> wrote: > >>>>> > >>>>> 2017-05-02 18:20 GMT+03:00 John Erling Blad <jeblad(a)gmail.com>om>: > >>>>>> > >>>>>> Brute force solution; turn the ContentTranslation off. Really > >>>>>>> > >>>>>> stupid > >>> > >>>> solution. > >>>>>>> > >>>>>> > >>>>>> ... Then I guess you don't mind that I'm changing the thread

name

:) > >>>>>> > >>>>>> > >>>>>> The next solution; turn the Yandex engine off. That would solve

> >>>>>>> part of the problem. Kind of lousy solution though. > >>>>>>> > >>>>>>> What about adding a language model that warns when the language > >>>>>>> > >>>>>> constructs > >>>>>> > >>>>>>> gets to weird? It is like a "test" for the translation. The CT

> >>>>>>> > >>>>>> used > >>>> > >>>>> for > >>>>>> > >>>>>>> creating a translation, but the language model is used for > >>>>>>> > >>>>>> verifying > >>> > >>>> if > >>>> > >>>>> the > >>>>>> > >>>>>>> translation is good enough. If it does not validate against the > >>>>>>> > >>>>>> language > >>>>> > >>>>>> model it should simply not be published to the main name space.

> >>>>>>> > >>>>>> will > >>>> > >>>>> still be possible to create a draft, but then the user is > >>>>>>> > >>>>>> completely > >>> > >>>> aware > >>>>>> > >>>>>>> that the translation isn't good enough. > >>>>>>> > >>>>>>> Such a language model should be available as a test for any > >>>>>>> > >>>>>> article, > >>> > >>>> as > >>>> > >>>>> it > >>>>>> > >>>>>>> can be used as a quality measure for the article. It is really

> >>>>>>> > >>>>>> quantity > >>>>> > >>>>>> measure for the well-spokenness of the article, but that isn't > >>>>>>> > >>>>>> quite > >>> > >>>> so > >>>> > >>>>> intuitive. > >>>>>>> > >>>>>>> So, I'll allow myself to guess that you are talking about one > >>>>>> > >>>>> particular > >>>> > >>>>> language, probably Norwegian. > >>>>>> > >>>>>> Several technical facts: > >>>>>> > >>>>>> 1. In the past there were several cases in which translators to > >>>>>> > >>>>> different > >>>> > >>>>> languages who reported common translation mistakes to me. I

passed

>>>>>> >>>>> them >>> >>>> on >>>>> >>>>>> to Yandex developers, with whom I communicate quite regularly.

They

>>>>>> acknowledged receiving all of them. I am aware of at least one

such > >>>>>> > >>>>> common > >>>>> > >>>>>> mistake that was fixed; possibly there were more. If you can

give

>>>>>> >>>>> a >>> >>>> list >>>>> >>>>>> of such mistakes for Norwegian, I'll be very happy to pass them

on.

I >>>>>> absolutely cannot promise that they will be fixed upstream, but

it's > >>>>>> possible. > >>>>>> > >>>>>> 2. In Norwegian, Apertium is used for translating between the

two

>>>>> >>>> varieties >>>> >>>>> of Norwegian itself (Bokmål and Nynorsk), and from other

Scandinavian >>>>>> languages. That's probably why it works so well—they are similar

>>>>> grammar, vocabulary, and narrative style (I'll pass it on to

Apertium >>>>>> developers—I'm sure they'll be happy to hear it). Unfortunately, >>>>>> >>>>> machine >>>> >>>>> translation from English is not available in Apertium. Apertium

works > >>>>>> > >>>>> best > >>>>> > >>>>>> with very similar languages, and English has two

characteristics,

> >>>>>> > >>>>> which > >>> > >>>> are > >>>>> > >>>>>> unfortunate when combined: it is both the most popular source

for

> >>>>>> translation into almost all other languages (including

Norwegian),

> >>>>>> > >>>>> and > >>> > >>>> it > >>>> > >>>>> is not _very_ similar to any other languages (except maybe

Scots).

>>>>>> >>>>> Machine >>>>> >>>>>> translation from English into Norwegian is only possible with

Yandex > >>>>>> > >>>>> at > >>> > >>>> the > >>>>> > >>>>>> moment. More engines may be added in the future, but at the

moment

>>>>>> >>>>> that's >>>> >>>>> all we have. That's why disabling Yandex completely would indeed

be a > >>>>>> > >>>>> lousy > >>>>> > >>>>>> solution: A lot of people say that without machine translation > >>>>>> > >>>>> integration > >>>>> > >>>>>> Content Translation is useless. Not all users think like that,

but

> >>>>>> > >>>>> many > >>> > >>>> do. > >>>>> > >>>>>> 3. We can define a numerical threshold of acceptable percentage

> >>>>>> > >>>>> machine > >>>>> > >>>>>> translation post-editing. Currently it's 75%. It's a tad > >>>>>> > >>>>> embarrassing, > >>> > >>>> but > >>>>> > >>>>>> it's hard-coded at the moment, but it can be very easily be made > >>>>>> > >>>>> into a > >>> > >>>> variable per language. If the translator tries to publish a page

> >>>>>> > >>>>> which > >>>> > >>>>> less than that is modified, a warning will be shown. > >>>>>> > >>>>>> 4. I'm not sure what do you mean by "language model". If it's

any

>>>>> >>>> kind >> >>> of a >>>> >>>>> linguistic engine, then it's definitely not within the resources

that >>>>>> >>>>> the >>>> >>>>> Language team itself can currently dedicate. However, if somebody

who

would > >>>>>> basically be an upgraded version of suggestion #1 above. (In my > spare > >>>>>> > >>>>> time > >>>>> > >>>>>> as a volunteer I'm doing something comparable for Hebrew,

although

>>>>> >>>> not >> >>> for >>>> >>>>> translation, but for improving how MediaWiki link trails work.) >>>>> _______________________________________________ >>>>> Wikimedia-l mailing list, guidelines at:

unsubscribe>

>>>>>> >>>>> _______________________________________________ >>>>> Wikimedia-l mailing list, guidelines at:

https://meta.wikimedia.org/

>>>> wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ >>>> wiki/Wikimedia-l >>>> New messages to: Wikimedia-l(a)lists.wikimedia.org >>>> Unsubscribe: https://lists.wikimedia.org/

mailman/listinfo/wikimedia-l, >>>>> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=

unsubscribe>

>>>>> >>>> _______________________________________________ >>>> Wikimedia-l mailing list, guidelines at:

https://meta.wikimedia.org/

>>>> wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ >>>> wiki/Wikimedia-l >>>> New messages to: Wikimedia-l(a)lists.wikimedia.org >>>> Unsubscribe: https://lists.wikimedia.org/

mailman/listinfo/wikimedia-l

, >>>> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=

unsubscribe> > >>>> > >>>> _______________________________________________ > >>> Wikimedia-l mailing list, guidelines at:

https://meta.wikimedia.org/

>>> wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/ >>> wiki/Wikimedia-l >>> New messages to: Wikimedia-l(a)lists.wikimedia.org >>> Unsubscribe: https://lists.wikimedia.org/

mailman/listinfo/wikimedia-l, > >>> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=

unsubscribe>

>>> >>> _______________________________________________ >> Wikimedia-l mailing list, guidelines at:

https://meta.wikimedia.org/wik > >> i/Mailing_lists/Guidelines and https://meta.wikimedia.org/wik > >> i/Wikimedia-l > >> New messages to: Wikimedia-l(a)lists.wikimedia.org > >> Unsubscribe: https://lists.wikimedia.org/

mailman/listinfo/wikimedia-l

, > >> <mailto:wikimedia-l-request@lists.wikimedia.org?subject=

unsubscribe>

>> > > > > _______________________________________________ > Wikimedia-l mailing list, guidelines at:

https://meta.wikimedia.org/wik > > i/Mailing_lists/Guidelines and https://meta.wikimedia.org/ > wiki/Wikimedia-l > > New messages to: Wikimedia-l(a)lists.wikimedia.org > > Unsubscribe: https://lists.wikimedia.org/

mailman/listinfo/wikimedia-l,

<mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

John Erling Blad

11:20 a.m.

Note that some language pairs could easily be 100% correct. On Wed, May 3, 2017 at 1:06 PM, David Cuenca Tudela <dacuetu(a)gmail.com> wrote:

...

It does depend a lot on the engagement level of the human behind the keyboard. When I deal with machine-translated text, I simply wonder

whether

the someone behind the keyboard took efforts to actually read the piece. Now whether this would work if limited to namespaces outside "main" - I

space

translator.

>> >> When the tool is used in the right way, it is a fantastic tool. >> >> Maybe we can find better methods to nudge people toward taking their

time

articles.

"Oppland

>>>> >>> er >> >>> et fylke i…". >>>> >>>> (I just checked and it seems like Yandex messes up a lot less now

than

>>>> previously, but it is still pretty bad.) >>>> >>>> Apertium works because the language is closely related, Yandex does

not

>>>> work because it is used between very different languages. People try

number

then

>>>> >>> the >>> >>>> sentence must be checked. It is not always wrong to write "en jenta"

>>>> Norwegian, but it is likely to be wrong. >>>> >>>> A language model could be a statistical model for the language

itself,

>>>> >>> not >>> >>>> for the translation into that language. We don't want a perfect >>>> >>> language >> >>> model, but a sufficient language model to mark weird constructs. A

very

text,

Scandinavian

>>>>> languages. That's probably why it works so well—they are similar in >>>>> grammar, vocabulary, and narrative style (I'll pass it on to

Apertium

that

spare

https://meta.wikimedia.org/

mailman/listinfo/wikimedia-l,

Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wik i/Mailing_lists/Guidelines and https://meta.wikimedia.org/wik i/Wikimedia-l New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

_______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wik i/Mailing_lists/Guidelines and https://meta.wikimedia.org/

wiki/Wikimedia-l

New messages to: Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

Amir E. Aharoni

11:45 a.m.

2017-05-03 14:06 GMT+03:00 David Cuenca Tudela <dacuetu(a)gmail.com>om>:

...

As I noted, this already exists. Set at 75%. Can be changed. -- Amir Elisha Aharoni · אָמִיר אֱלִישָׁע אַהֲרוֹנִי http://aharoni.wordpress.com ‪“We're living in pieces, I want to live in peace.” – T. Moore‬

Yaroslav Blanter

6:46 p.m.

My idea was that adding an extra button to press brings the probability of the whole process down. If someone is determine to systemically add bad machine translations to the main namespace I guess only blocks could help. On the other hand, and extra button gives at leat an opportunity to read the result and reflect on it. Cheers Yaroslav

2587

days inactive

2588

days old

wikimedia-l@lists.wikimedia.org

Manage subscription

15 comments

9 participants

tags (0)

participants (9)

Amir E. Aharoni
Bodhisattwa Mandal
David Cuenca Tudela
John Erling Blad
Lodewijk
Pharos
Wojciech Pędzich
Yaroslav Blanter
Ziko van Dijk