CafeTran, Machine Translation and auto-assemble

[abola...@icab.cat]

Hello,

 

I am new to CafeTran but not to CATs. I am subscriber to DeepL, which I have managed to enable on CafeTran. 

 

 

It often happens to me that I am not satisfied with certain term translations that I obtain with DeepL. For example, (assuming an ENG>Spa project) DeepL almost invariably yields "Acuerdo" where I want "Contrato".

 

Now ,I am aware that by ticking the box "Team auto-assembling with machine translation" under Edit>Preferences>Machine Translation, in theory I could get “Contrato” (a term in my glossary) to pop up instead of Acuerdo (DeepL). The feature works sometimes, but more often than not, it doesn't so instead of the relevant word (Contrato), DeepL produces a string of capital letters, ZZZZXXXXYYYY. 

 

So an entire paragraph translated by DeepL would look like this:

 

El ZZZYYYYXXX no tendrá derecho a reclamar, en virtud de la presente VVVUUUTTT, indemnización alguna, costos de oportunidad o lucro cesante, o cualquier otra pérdida o daño sufrido por el ZZZYYYYYXXX que no sea el NNNMMMLLLLLKKKJJJJI

 

Where each string of capital letters corresponds to a word which DeepL “knows” I have told it to replace with another but, instead of the word I want, it gives me ZZZYYYYXXX. 

 

Does anyone know why this happens? And more importantly, how to address this?

 

I will be very grateful if someone can shed some light on this.

 

Thank you!

Alex

[Igor Kmitowski]

Hello Alex,

Try switching off "Masking of non-translatable fragment" in Edit > Preferences > Machine Translation.

Igor

[abola...@icab.cat]

Hi Igor,

Thank you for your super fast response. This option is already unticked. Any other clues'

Thank you again.

Alex

[Igor Kmitowski]

If you just unchecked it, please restart the program. If it doesn't help, you can update CafeTran the latest version as described here

Make sure you complete all your unfinished projects before updating. If it still doesn't help, reset the preferences via Edit > Preferences > the Reset button.

Igor

--
You received this message because you are subscribed to the Google Groups "CafeTranslators" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cafetranslato...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Alex Boladeras]

Hi Igor,

 

Thank you for your help here. I followed your suggested steps (including purchasing a full license :) but they did not work.:

 

1 Buying license

2 Installing the newest update

3 Resetting the MT preferences

4 Rebooting Memotran

 

Can you suggest more possible solutions? Being able to integrate my MT service (DeepL) with my local TM and Glossaries via auto-assemble is crucial for me and, I guess, for many other translators too.

 

Thank you for your support.

 

Best wishes,

 

Alex Boladeras

Spanish lawyer

Solicitor admitted in England and Wales

Legal Translator NAATI 66594 (Australia), DipTrans (UK)  
 

m +62 (0) 87862150423 
e  abola...@icab.cat 

Skype name: bula1404

 

CONFIDENTIALITY NOTICE This message contains confidential information, which might also be legally privileged, and it is addressed exclusively to its intended recipient. If you have received it in error, please notify us immediately by replying to this email and then delete it from your system.You are hereby informed that any non authorized use, copy or disclosure of the contents of this message to third parties is strictly forbidden. Email and Internet do not guarantee the confidentiality and/or security of communications. Therefore, we disclaim any liability in this respect. Be green. Keep it on the screen!

[Igor Kmitowski]

Hello Alex,

There is a new update available with the slightly changed algorithm in this feature. Please check here for details:

Thanks,

Igor

[Alex Boladeras]

Hi Igor,

Yes, that’s the update I referred to in my previous email and still not good. Terms stored in my TM or Glossaries still appear as XXXXYYYYYZZZZ with DeepL. 

Thank you for your support in helping implement this (vital) feature.

Best,

Sent from my iPhone

Alex Boladeras

[Igor Kmitowski]

It's a new build of the same update released only today. Just repeat the updating procedure (https://cafetran.freshdesk.com/discussions/topics/6000055053)

and you should notice the difference.

Igor

[Alex Boladeras]

Hi Igor,

 

Let my start by saying that I am deeply grateful for your support with this issue.

 

The latest update is definitely an improvement but we are not quite there yet.

 

The good news is that I don’t get the XXXXYYYYY string of letters.

 

Instead what I get now is this:

 

ISSUE 1

 

On the lower left pane (I think it’s called Resources) I get {1}, {2} etc where the number is the word stored on my Glossary. Sometimes there are as many as eight {} or more.

 

On the lower right pane (Results?) the {1] [2] etc. are usually (but not always) converted into words. So occasionally, I might still get the occasional {1}

 

ISSUE 2

 

Word substitutions (ie my Glossary/TM entry replacing the DeepL term) are not perfect.

 

In one segment, replacements occur perfectly (ie autoassembly replacing my glossary/TM entries with those provided by DeepL), but the next segment (for no apparent reason) some terms are simply not replaced.

 

ISSUE 3

 

Occasionally, what I also do get are two terms completely pertaining to two different entries in my Glossary.

So, let´s say in my Glossary I have Entries 1, 2, 3 etc. Let’s also assume that my source text contains entry 1 only. Where DeepL and autoassembly work together, I get entry 1 and, just beside it, entry 5 of my glossary, instead of entry 1 only.

 

ISSUE 4: Conflict TM and Glossary

 

Scenario: TM contains a fragment in ENG and SPA

Glossary conains a term in ENG and SPA

 

DeepL and autoassemble work to yield the TM fragment but I would prefer the Glossary. Is there a way to resolve this issue?

 

I think that’s all for the moment.

Thank you again for your invaluable and top-notch support.

 

Kind regards,

Alex

[Igor Kmitowski]

Hi Alex,

There is a similar thread here so I answer there avoiding cross-posting.

Igor

[Alex Boladeras]

Thank you so much, Igor.

 

I will continue posting my observations, remarks, etc on that thread.

 

BTW: you may or may not know that MemoQ’s latest version does NOT include this feature yet which I find crucial in order to make the most of MT services. So hats off to you...

 

Best wishes,

 

Alex

[Alex Boladeras]

Hi again, Igor

 

Sorry to trouble you again but I can log in to the forum with my credentials. And when I click Resend Password, no email arrives to my Inbox. This is odd as I have no problem in logging into the home site...

 

Do you think you could help me with that?

 

Thank you.

 

Alex Boladeras

Spanish lawyer

Solicitor admitted in England and Wales

Legal Translator NAATI 66594 (Australia), DipTrans (UK)  
 

m +62 (0) 87862150423 
e  abola...@icab.cat 

Skype name: bula1404

 

CONFIDENTIALITY NOTICE This message contains confidential information, which might also be legally privileged, and it is addressed exclusively to its intended recipient. If you have received it in error, please notify us immediately by replying to this email and then delete it from your system.You are hereby informed that any non authorized use, copy or disclosure of the contents of this message to third parties is strictly forbidden. Email and Internet do not guarantee the confidentiality and/or security of communications. Therefore, we disclaim any liability in this respect. Be green. Keep it on the screen!

 

From: <cafetra...@googlegroups.com> on behalf of Igor Kmitowski <cafetran...@gmail.com>

Reply-To: <cafetra...@googlegroups.com>
Date: Monday, 27 August 2018 at 16:19