Google Blogoscoped

Forum

Google Translator Toolkit  (View post)

TOMHTML [PersonRank 10]

Tuesday, June 9, 2009
15 years ago17,469 views

http://translate.google.com/toolkit

it's a tool that allows you to translate files and webpages, and **edit** the translation (in wiki-style, multi-users).
http://i40.tinypic.com/e7kt9j.jpg

Tony Ruscoe [PersonRank 10]

15 years ago #

Very interesting! Especially since SDL (the company I work for) just released the latest verson of their translation software (SDL Trados Studio 2009) last Friday. I'll be taking a very close look at this...

Tony Ruscoe [PersonRank 10]

15 years ago #

Some interesting bits from the support and Terms of Service:

<< When you upload a document into Google Translator Toolkit, we automatically 'pretranslate' your document as follows:

   1. We divide your document into segments, usually sentences, headers, or bullets.
   2. We search all available translation databases for previous human translations of each segment.
   3. If any previous human translations of the segment exist, we pick the highest-ranked search result and 'pretranslate' the segment with that translation.
   4. If no previous human translation of the segment exists, we use machine translation to produce an 'automatic translation' for the segment, without intervention from human translators. >>

http://translate.google.com/support/toolkit/bin/answer.py?hl=en&answer=147836

Assuming they mean *anyone's* human translation database (AKA translation memories), there's nothing in the Terms to say that Google can use your TMs like this:

<< Sharing your Translation Memory(-ies) with others and terminating your Google account

You can share your Translation Memory(-ies) with others, and you can revoke sharing privileges at any time. If you do not want your Translation Memory(-ies) to continue to be shared with the people you designated for sharing after you terminate your Google account, you must revoke sharing privileges prior to terminating your Google account. >>

http://translate.google.com/toolkit/TOS.html

[Edit]

Actually, this might cover that:

<< Use of your Content

By submitting your content through the Service, you grant Google the permission to use your content permanently to promote, improve or offer the Services. If Google publicly displays any of the content you submitted through the Service, Google will display only portion(s) and not the entirety of the content at one time. >>

I'd be interested to know whether that's correct. Surely if the document is the same, it could easily display the entirety of the content.

Cookie Lee [PersonRank 9]

15 years ago #

This seems to be previously known as Google Translation Center?
http://blogoscoped.com/archive/2008-08-04-n48.html

The following URL leads you to this service too.
https://www.google.com/accounts/ServiceLogin?service=gtrans

Above 4 comments were made in the forum before this was blogged,

hebbet [PersonRank 10]

15 years ago #

on Google Account it is called Google Translation Center

http://i39.tinypic.com/1pw2kg.png

Barnaby Capel-Dunn [PersonRank 2]

15 years ago #

Tony,
The application seems to insist on English as the source language. Have you encountered this problem?

Tony Ruscoe [PersonRank 10]

15 years ago #

Barnaby, that's a good point! That does seem to be the case. Even uploading or selecting a non-English document doesn't allow you to change the source language.

Barnaby Capel-Dunn [PersonRank 2]

15 years ago #

It's not just me, then! If they can sort out this problem, it looks interesting, though.

高德翔 [PersonRank 0]

15 years ago #

这玩意太彪悍了,译言可以直接砍掉一半了。

Barnaby Capel-Dunn [PersonRank 2]

15 years ago #

Tony, here is the explanation (in "Help")

For now, Google Translator Toolkit supports the following languages:

Source languages: English
Target languages: Albanian, Arabic, Bulgarian, Catalan, Chinese (Simplified), Chinese (Traditional), Croatian, Czech, Danish, Dutch, English, Estonian, Filipino, Finnish, French, Galician, German, Greek, Hebrew, Hindi, Hungarian, Indonesian, Italian, Japanese, Kannada, Korean, Latvian, Lithuanian, Malayalam, Maltese, Marathi, Norwegian (Bokmal), Polish, Portuguese (Brazil), Portuguese (Portugal), Romanian, Russian, Serbian, Slovak, Slovenian, Spanish, Swedish, Tamil, Telugu, Thai, Turkish, Ukrainian, and Vietnamese

Tony Ruscoe [PersonRank 10]

15 years ago #

Barnaby, thanks for the info. Here's the link for reference:
http://translate.google.com/support/toolkit/bin/answer.py?hl=en&answer=147837

Tony Ruscoe [PersonRank 10]

15 years ago #

Here's the Official Google Blog post:

http://3.bp.blogspot.com/_7ZYqYi4xigk/Si5vvl2QzXI/AAAAAAAAD4Q/tFw79CJn-ZI/s400/cycle.PNG

http://googleblog.blogspot.com/2009/06/translating-worlds-information-with.html

Li Na [PersonRank 0]

15 years ago #

Translation memory applications are nothing new, but the Knol and Wikipedia additions are really innovative. As a freelance translator myself – who uses translation memory tools for 3 years already (I work in OneHourTranslation.com and translation agencies) I will give this toolkit a try.

WebSonic.nl [PersonRank 10]

15 years ago #

http://www.youtube.com/watch?v=C7W2NJFdoIg

Ian Davies [PersonRank 2]

15 years ago #

Um. I was pretty excited about this, but why is English the only source language available? Almost <em>all</em> of my translation is <em>into</em> English from other languages! And why can't we upload .docx files? You can into Google Docs now, and finding a file on my computer both in .doc format and in English to test this thing was near impossible!

If it takes off, it's certainly going to be fantastic data for Google to use to improve their machine translation, with all those parallel texts to analyse. It'll have to be made pretty compelling for translators to want to switch to it – it's very spartan at the moment. The dictionary is extremely basic too, not even showing the genders next to nouns, let alone offering any sort of verb conjugator or thesaurus, and again, not letting you select the language. On the plus side, I'm glad that you can stop it filling in the right pane with the machine translation.

The translation memory & glossary could end up being brilliant if enough people contribute, but for now I've got my own tools for that. I hope Google improve on this, because if made into a good tool it's something I would use every day. The ironic thing is that by using this, translators are contributing to their future unemployment if the machine translation becomes intelligent enough as a result! That's certainly a long way off, but just imagine the applications – the Google Wave translation demo left me in awe of the future possibilities.

Trogdor [PersonRank 6]

15 years ago #

"There are literally thousands of Wikipedia articles being translated all the time, but the translations aren't usually maintained in a translation memory."

Another problem is "translated" Wikipedia articles are often not exact translations of eachother – often one language will be comprehensive, and others will be basic, or incomplete "stubs" waiting for someone to come along and improve them. Other times, the articles will simply have a different set of headings, content, even facts (!) in other languages.

Andy Wong [PersonRank 10]

15 years ago #

Apart from other similar products existing, Google's translation has much better collective intelligence. Translation and collective intelligence benefit each other.

Jack Volks [PersonRank 0]

15 years ago #

Tony, thanks for the interesting article.
I hope you don't mind me asking you, as an SDL employee, how your company views this Google service, which, if it takes off, could become very serious competition for SDL. Or is Google licensing SDL CAT technology to operate the service?

Barnaby Capel-Dunn [PersonRank 2]

15 years ago #

Tony,
Pending the arrival of more source languages in Google Translator Toolkit, here's a very useful, if more limited, workaround:
http://googletranslateclient.com/
It saves quite a bit of time.
Best
Barnaby

Tony Ruscoe [PersonRank 10]

15 years ago #

> I hope you don't mind me asking you, as an SDL employee,
> how your company views this Google service, which, if it takes
> off, could become very serious competition for SDL. Or is
> Google licensing SDL CAT technology to operate the service?

Jack, I can't really answer that. Sorry.

Martin Wunderlich [PersonRank 0]

15 years ago #

I wonder, how will own the copyright to the translations in google's massive TM. Will the translators be asked, if they waive the copyright so that anyone can use their translations elsewhere?

These reservations aside, I think it's a great idea to build this massive global TM – even if google's main motivation might be to get structured data for their MT system. The next thing should be standardised API for accessing this TM, using something like globalsight's API (http://www.globalsight.com). This should be open and standardised through an organisation like OASIS.

Cheers,

Martin

Guillaume de Brébisson [PersonRank 0]

15 years ago #

Hello everybody,

I don't know what to think about the Google initiative: being the owner of a translation company (much smaller than SDL), I dislike having such a new competitor to face with.

Now, from a technological point of view, I am sure that combining translation engines, translation memories and human translation know-how is valuable and efficient. I experienced this in the past, using a traditional rule based translation engine, and it proved to be highly productive. That is also why I published a review of Systran's last version in my blog, L'Observatoire de la traduction (blog.anyword.fr).

However, I can see a major difference between both products: Google uses a statistical translation engine when Systran uses a hybrid engine (rules based and statistic), and Google works on a huge non-differenciated corpus (all the translation memories stored by the translators) when Systran trains its engine on domain-centric, specific, user-defined corpus.

In my opinion, Systran should deliver better results than Google. But it's not free... What do you think about all this, anyway?

All the best,
Guillaume de Brébisson

[Signature URL removed – Tony]

UNL [PersonRank 0]

15 years ago #

Hello, and sorry about my English.

I think a god aid to on-line translate is the UNL (Universal Networking Language).
http://en.wikipedia.org/wiki/Universal_Networking_Language
http://www.vai.dia.fi.upm.es/ing/projects/unl/index.htm

I think this is a good project, but there is stopped or slower advance. If google take it for google translator, an advance on on-line translations can arrive soon.

Stephen [PersonRank 0]

15 years ago #

[put at-character here] Guillaume
Google's MT may be trained on a global base of content, but Google is excellent at drawing correlations between documents, classifying documents etc. They could use this knowledge to give more domain focused translation.

[put at-character here] Tony
You think Google might have launched this at the same time as SDL and Alchemy on purpose? Translators may hold off shelling out for a desktop product until they see what Google might offer. It ain't half as fully featured as the state-of-the-art CAT tools, but with Google's engineering power there is no reason it couldn't be in the future. The bottom line is they want to attract as many translators as possible to get as much data as possible.

nunobaba [PersonRank 0]

15 years ago #

The same service already exists for almost 2 years on the web. Traduwiki.org ([traduwiki.org]) is a translation tool designed for group and organization purposes. Functionalities are the same. Google isn't innovating on this one.

bumblefish [PersonRank 0]

15 years ago #

Social translation tools like jollo.com have existed for a long time.

[Unlinked – Tony]

Forum home

Advertisement

 
Blog  |  Forum     more >> Archive | Feed | Google's blogs | About
Advertisement

 

This site unofficially covers Google™ and more with some rights reserved. Join our forum!