Help with Machine Translation in Déjà Vu. Thread poster: Gabriel Araújo
|
Hi! Here’s what we are doing (and is resulting the less issues for us …): First round we use "Accept only exact matches” Afterwards, in sequence, we run activating MT (Use Machine Translation for whole segments) and keeping "Assemble from Portions” disabled NOTE: If we activate "Assemble from Portions" DVX3 will translate word by word (no context considered…) generating a useless result ... With that we got some nice transla... See more Hi! Here’s what we are doing (and is resulting the less issues for us …): First round we use "Accept only exact matches” Afterwards, in sequence, we run activating MT (Use Machine Translation for whole segments) and keeping "Assemble from Portions” disabled NOTE: If we activate "Assemble from Portions" DVX3 will translate word by word (no context considered…) generating a useless result ... With that we got some nice translations, but we get issues basically with 2 points: 1- It’s not taken in consideration our Glossary of terms (Term Base) 2- Sentences with tags (see below) are not properly translated (basically it forces the words between the tags to be translated separately …) Example: a)The {15}Wall{16} tool is automatically selected and the selected wall style is Applied to the wall as t is drawn. b) The Wall tool is automatically selected and the selected wall style is Applied to the wall as it is drawn. In “a” above the result for Wall too is incorrect (because of the tag {16}) In “b” (in which we removed the tag in the same sequence…) the resulted translation is fine...) So my 3 questions are: 1) Can we ask Machine Translation (in our case, we are using the Google Translator API…) to use our Termbase? 2) Is there any way to ask DVX3 to ignore (temporarily filter out…) the tags for a clear usage of the MT? 3)If we manage to ignore the tags, will they be added in the correct places or will they have a bug? We are not using the Lexicon … is this (maybe) an alternative to solve this? ▲ Collapse | | | Fine-tune Deja Vu options? | Oct 13, 2021 |
Have you ticked 'Options > Translation > Assemble options > Use Machine Translation'? If you have, you could try using 'Assemble from Portions' as an alternative to 'Use Machine Translation for whole segments'. This might improve consistency with your termbase. A side effect would be more accurate handling of tagged portions provided their reference translations have been saved to your TM, termbase or lexicon. To answer your questions: 1. Can we ask Machine T... See more Have you ticked 'Options > Translation > Assemble options > Use Machine Translation'? If you have, you could try using 'Assemble from Portions' as an alternative to 'Use Machine Translation for whole segments'. This might improve consistency with your termbase. A side effect would be more accurate handling of tagged portions provided their reference translations have been saved to your TM, termbase or lexicon. To answer your questions: 1. Can we ask Machine Translation (in our case, we are using the Google Translator API…) to use our Termbase? 2. Is there any way to ask DVX3 to ignore (temporarily filter out…) the tags for a clear usage of the MT? I'm not sure, either can be done if you use MT to translate whole segments.
[Edited at 2021-10-13 18:54 GMT] ▲ Collapse | | | Endre Both Germany Local time: 10:55 English to German Use other MT providers; remove tags yourself | Oct 13, 2021 |
a) There are MT providers that take glossaries into account (DeepL, Systran and even some Google offerings, though I'm not sure you can use those in DVX). b) You cannot change how MT works in DVX, but if you're dealing with reasonably large volumes that justify spending some time on optimized MT, you can make a copy of the project, remove all tags from the source segments via SQL or search and replace, and pretranslate the tagless segments. Then via SQL or copy-and-paste in MS Acces... See more a) There are MT providers that take glossaries into account (DeepL, Systran and even some Google offerings, though I'm not sure you can use those in DVX). b) You cannot change how MT works in DVX, but if you're dealing with reasonably large volumes that justify spending some time on optimized MT, you can make a copy of the project, remove all tags from the source segments via SQL or search and replace, and pretranslate the tagless segments. Then via SQL or copy-and-paste in MS Acces, you copy over the MT'd segments to the original project, adding back the tags as appropriate during the manual translation stage. ▲ Collapse | | | We can't use MT with TM? | Oct 13, 2021 |
So the best way to use MT is it really on untagged lines or Pre-Translate and Review manually? Thanks for all the help. | | | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Help with Machine Translation in Déjà Vu. Wordfast Pro | Translation Memory Software for Any Platform
Exclusive discount for ProZ.com users!
Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value
Buy now! » |
| CafeTran Espresso | You've never met a CAT tool this clever!
Translate faster & easier, using a sophisticated CAT tool built by a translator / developer.
Accept jobs from clients who use Trados, MemoQ, Wordfast & major CAT tools.
Download and start using CafeTran Espresso -- for free
Buy now! » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |