Re: [Vo]:Google translation of an electrochemistry paper

[ROGER ANDERTON]

world's go mad re: what's happening in US politics

anyway see my unified field theory website:
http://unifiedfieldtheory.co.uk/

I got the unified field theory it was published in 1758 and others have worked 
along similar lines; a tradition of physics that's different to mainstream
also my  latest paper: "they" misunderstood Einstein's relativity it got 
mistranslated from the original German of 1905 when translated into 
Englishhttp://gsjournal.net/Science-Journals/Research%20Papers-Relativity%20Theory/Download/6892


 

On Friday, 21 April 2017, 19:19, Daniel Rocha  wrote:
 

 Translations from and to English are good. Not so much to other languages.In 
any case, I also cheat :-)

One thing I miss it is that they should provide an OCR, so that I could read 
letters in photographs, graphs and comics (korean, japanese, chinese, french or 
whatever). It would also boost people skills in any language, since you'd have 
the object in context.
2017-04-21 13:04 GMT-03:00 Jed Rothwell :

A researcher asked me to translate two papers from Japanese into English:

https://www.jstage.jst.go.jp/ article/jsms1963/49/11/49_11_ 1242/_article
https://www.jstage.jst.go.jp/ article/jsms1963/50/9/50_9_ 999/_article


The first one is:

(J. Soc. Mat. Sci., Japan), Vol.49, No. 11, pp. 1242-1248, Nov. 2000
Deformation and Aging of Pd by Hydrogen Absorption-Desorption Cycles
— Deformation of Pd at a Hydrogen Absorption-Desorption Cycle —

Young-guan JUNG, Hideki SUEHIRO and Yuzuru SAKAI


I use Google translate to do this. (https://translate.google.com/ ) It feels 
like cheating, but is sure makes things easier. However, you cannot simply dump 
an Acrobat document into Google translate. That does not work well. I thought I 
would list some of the steps you should to take to make Google translate work, 
and show an example of how the text comes out.
Google translate makes mistakes, as shown below, but in the last few months it 
has improved a great deal thanks to the newest neural network AI techniques. 
See:

https://blog.google/products/ translate/found-translation- 
more-accurate-fluent- sentences-google-translate/


Here is the worst mistake in the first paragraph:

. . . the occurrence of microcracks and its progress process, many problems, 
durability of the hydrogen storage material It is an obstacle to improve sex."

The last part should be:

. . . These problems are obstacles to improving the durability of the hydrogen 
storage material."

Nothing to do with sex! Perhaps people often ask Google translate to translate 
documents about sex, so the AI thinks everything is about sex.

(Oddly enough, that mistake goes away when you submit the sentence fragment on 
its own, outside of the paragraph.)


Anyway, to translate an Acrobat document from Japanese into English, take the 
following steps:
1. Convert the text to Microsoft Word, using a program such as Power PDF. This 
is essential, mainly because it preserves most paragraphs. Submitting the 
Acrobat document as is, or copying the text will produce many errors. Every 
line in a paragraph will break, which will produce nonsensical translations.
2. Fix paragraphs broken by figures and the pages.
3. Eliminate multiple columns, figures, and all of the formatting you can.
4. Submit the text to Google translate.
5. Compare the resulting text to the Japanese original. It is very handy to use 
a voice reading program such as TextAloud (http://nextup.com/) to read the text 
in Japanese as you look through the English text, and vice versa.
6. Correct and adjust the text.
Google translate will often select words that are correct and understandable, 
but they may not be what is normally used in this context. For example, it 
translated Japanese term "suiso kyuuzou" as "hydrogen occlusion." I think 
electrochemists usually say "absorption." Both terms are listed in a dictionary:
水素吸蔵    
   - hydrogen absorption
   - hydrogen input
   - hydrogen occlusion


Okay. Here is the Google version of the entire first paragraph with no changes:

Regarding the behavior of hydrogen in the metallic structure, many studies 1) - 
3) have been done mainly concerning the hydrogen embrittlement problem. Solid 
dissolved hydrogen is trapped in dislocations, voids and the like in a steel 
material structure such as carbon steel and stainless steel, and is thought to 
be a factor that promotes destruction, and researches on elucidation of the 
material embrittlement mechanism by hydrogen are being conducted. On the other 
hand, recently, from the viewpoint of global environmental problems, 
development of a hydrogen storage material as a clean hydrogen energy carrier 
is actively underway, that is, some metals including rare earth metals easily 
form hydride It has the ability to absorb and release about 1000 times as much 
hydrogen as its own volume. As already seen in nickel-metal hydride batteries 
5), etc., this product has been commercialized 

Re: [Vo]:Google translation of an electrochemistry paper

[Daniel Rocha]

Translations from and to English are good. Not so much to other
languages.In any case, I also cheat :-)

One thing I miss it is that they should provide an OCR, so that I could
read letters in photographs, graphs and comics (korean, japanese, chinese,
french or whatever). It would also boost people skills in any language,
since you'd have the object in context.

2017-04-21 13:04 GMT-03:00 Jed Rothwell :

> A researcher asked me to translate two papers from Japanese into English:
>
> https://www.jstage.jst.go.jp/article/jsms1963/49/11/49_11_1242/_article
> https://www.jstage.jst.go.jp/article/jsms1963/50/9/50_9_999/_article
>
>
> The first one is:
>
> (J. Soc. Mat. Sci., Japan), Vol.49, No. 11, pp. 1242-1248, Nov. 2000
> Deformation and Aging of Pd by Hydrogen Absorption-Desorption Cycles
> — Deformation of Pd at a Hydrogen Absorption-Desorption Cycle —
>
> Young-guan JUNG, Hideki SUEHIRO and Yuzuru SAKAI
>
>
> I use Google translate to do this. (https://translate.google.com/) It
> feels like cheating, but is sure makes things easier. However, you cannot
> simply dump an Acrobat document into Google translate. That does not work
> well. I thought I would list some of the steps you should to take to make
> Google translate work, and show an example of how the text comes out.
>
> Google translate makes mistakes, as shown below, but in the last few
> months it has improved a great deal thanks to the newest neural network AI
> techniques. See:
>
> https://blog.google/products/translate/found-translation-
> more-accurate-fluent-sentences-google-translate/
>
>
> Here is the worst mistake in the first paragraph:
>
> . . . the occurrence of microcracks and its progress process, many
> problems, durability of the hydrogen storage material It is an obstacle to
> improve sex."
>
> The last part should be:
>
> . . . These problems are obstacles to improving the durability of the
> hydrogen storage material."
>
> Nothing to do with sex! Perhaps people often ask Google translate to
> translate documents about sex, so the AI thinks everything is about sex.
>
> (Oddly enough, that mistake goes away when you submit the sentence
> fragment on its own, outside of the paragraph.)
>
>
> Anyway, to translate an Acrobat document from Japanese into English, take
> the following steps:
>
> 1. Convert the text to Microsoft Word, using a program such as Power PDF.
> This is essential, mainly because it preserves most paragraphs. Submitting
> the Acrobat document as is, or copying the text will produce many errors.
> Every line in a paragraph will break, which will produce nonsensical
> translations.
>
> 2. Fix paragraphs broken by figures and the pages.
>
> 3. Eliminate multiple columns, figures, and all of the formatting you can.
>
> 4. Submit the text to Google translate.
>
> 5. Compare the resulting text to the Japanese original. It is very handy
> to use a voice reading program such as TextAloud (http://nextup.com/) to
> read the text in Japanese as you look through the English text, and vice
> versa.
>
> 6. Correct and adjust the text.
>
> Google translate will often select words that are correct and
> understandable, but they may not be what is normally used in this context.
> For example, it translated Japanese term "suiso kyuuzou" as "hydrogen
> occlusion." I think electrochemists usually say "absorption." Both terms
> are listed in a dictionary:
>
> 水素吸蔵
> [image: 拡張検索] 
>
>- hydrogen absorption
>- hydrogen input
>- hydrogen occlusion
>
>
>
> Okay. Here is the Google version of the entire first paragraph with no
> changes:
>
> Regarding the behavior of hydrogen in the metallic structure, many studies
> 1) - 3) have been done mainly concerning the hydrogen embrittlement
> problem. Solid dissolved hydrogen is trapped in dislocations, voids and the
> like in a steel material structure such as carbon steel and stainless
> steel, and is thought to be a factor that promotes destruction, and
> researches on elucidation of the material embrittlement mechanism by
> hydrogen are being conducted. On the other hand, recently, from the
> viewpoint of global environmental problems, development of a hydrogen
> storage material as a clean hydrogen energy carrier is actively underway,
> that is, some metals including rare earth metals easily form hydride It has
> the ability to absorb and release about 1000 times as much hydrogen as its
> own volume. As already seen in nickel-metal hydride batteries 5), etc.,
> this product has been commercialized and the demand is rapidly increasing
> year by year. Furthermore, as hybrid cars that are collecting the topic of
> the future as a future model car also hydrogen batteries using misch metal
> are mounted, enabling environmentally friendly and fuel-efficient systems
> where metal materials occlude hydrogen, metal crystals As hydrogen enters
> the lattice, lattice expansion occurs, resulting in bulk expansion near