| Pages in topic: [1 2] > |
How to extract terminology from aligned documents using Multiterm? Thread poster: Ivan Chamme
|
|---|
Hello,
I have aligned several EN and ES Word documents, and I'd like to extract the bilingual terminology using Multiterm. I do not find MT options on the alignment interface. What options do I have?
I have also tried from the TM interface after importing the aligned segments into the selected TM.
Thank you!
| | | |
Stepan Konev 
Russian Federation
Local time: 09:30
English to Russian
| ProjectTerm Extract | Nov 15, 2022 |
I don't know how to do that with Multiterm, but you can use a plugin for that.
Try this video.
| | | |
Stepan Konev wrote: I don't know how to do that with Multiterm, but you can use a plugin for that. Try this video.
But you’ll have to translate the extracted terms manually. She didn’t even use the concordance…
See here for an alternative approach:
https://www.star-ts.com/translation/dynamic-linking/
[Edited at 2022-11-16 23:17 GMT]
| | | |
Samuel Murray
Netherlands
Local time: 08:30
Member (2006)
English to Afrikaans
+ ...
Ivan Chamme wrote:
I'd like to extract the bilingual terminology...
I don't think this is possible with Trados. I believe MemoQ does this on-the-fly (using available translation memories), but I don't think MemoQ allows you to save the terminology so extracted.
That video shows monolingual terminology extraction.
| | |
|
|
|
Stepan Konev
Russian Federation
Local time: 09:30
English to Russian
| RE: MemoQ does this on-the-fly | Nov 17, 2022 |
Samuel Murray wrote:
I believe MemoQ does this on-the-fly (using available translation memories), but I don't think MemoQ allows you to save the terminology so extracted. It does allow to save the terminology so extracted, but it won't translate it too. And it is quite expectable because memoQ can't say which target word is the equivalent of the source. It only inserts translations where the candidate term is already a true term.
| | | |
Stepan Konev
Russian Federation
Local time: 09:30
English to Russian
Hans Lenting wrote:
She didn’t even use the concordance… She didn't, but why can't you?
[Edited at 2022-11-17 10:59 GMT]
| | | |
Stepan Konev
Russian Federation
Local time: 09:30
English to Russian
Ok, I have read the article but I still don't understand how it (the DL mechanism) can judge which word in target segment is an equivalent term considering different syntax in different languages... The autosuggest engine processes longer phrases to make a suggestion—that is clear to me. But how can it parse single words? Probably I have to try it first to understand how it works.
| | | |
James Plastow
United Kingdom
Local time: 07:30
Member (2020)
Japanese to English
| manual but done as efficiently as possible | Nov 17, 2022 |
I would probably:
- make a 2 column table in Word with the source and target sentences
- highlight the source word and corresponding target word in each row manually (if there were two or more keywords in one sentence I would insert extra rows)
- use Word Advanced Find and Replace to delete everything other than the highlighted words
- convert that table of words to a termbase
I don't think there is any automated tool that would be able to do this.
| | |
|
|
|
Stepan Konev wrote:
Ok, I have read the article but I still don't understand how it (the DL mechanism) can judge which word in target segment is an equivalent term considering different syntax in different languages... The autosuggest engine processes longer phrases to make a suggestion—that is clear to me. But how can it parse single words? Probably I have to try it first to understand how it works.
I'm wondering whether you have discovered how DL works.
| | | |
Stepan Konev
Russian Federation
Local time: 09:30
English to Russian
| Secrets of DL operation | Dec 24, 2022 |
Hans Lenting wrote:
I'm wondering whether you have discovered how DL works. Is it a secret? The first sentence of the article reads: Dynamic Linking works like a multilingual concordance search. It displays all the segments which contain particular pairs of terms.
I mean it can be difficult to parse single words because of different syntax in different languages. For example, in Japanese they often omit pronouns. How can you parse a pair of terms when you have only one in your TM. The same applies to the Russian language. "Put your hands on your head" translates to "Руки за голову" (put and 2×your omitted).
However I admit I could misunderstand something. That's why I said I probably have to try it first [before I judge it].
[Edited at 2022-12-24 15:48 GMT]
| | | |
Samuel Murray
Netherlands
Local time: 08:30
Member (2006)
English to Afrikaans
+ ...
Stepan Konev wrote:
It does allow to save the terminology so extracted, but it won't translate it too. And it is quite expectable because memoQ can't say which target word is the equivalent of the source.
Okay, well, from my perspective, while translating, it certainly appears as if MemoQ is making guesses about what the possible translations of the terms can be (it shows this in the translation memory view pane), and sometimes the guesses are correct and sometimes they are wrong... and sometimes, as you say, MemoQ shows only the source text of a potential term, without a corresponding guess about what its translation might be.
This type of guessing would work better in some language pairs than in others. Both my source and target languages (English and Afrikaans) are from broadly similar branches of the language tree. The only difficulty with Afrikaans is that Afrikaans writes compound nouns as single words (e.g. "document alignment" is one word in Afrikaans, but it still consists of the stem "document" and the stem" alignment", so a clever tool might be able to run a partial dictionary attack on the target segment and determine what "document" and "alignment" are in Afrikaans, or what the Afrikaans for "file alignment" and "segment alignment" might be). And if a word in the English text is preceded by an article, odds are it would be preceded by an article in Afrikaans, too.
| | | |
Stepan Konev wrote:
Is it a secret? The first sentence of the article reads: Dynamic Linking works like a multilingual concordance search. It displays all the segments which contain particular pairs of terms.
Hahaha, secret? I meant the technology behind DL. Something like:
https://youtu.be/H-u_CfAuj2c
[Edited at 2022-12-24 18:09 GMT]
| | |
|
|
|
| |
Ivan Chamme
Local time: 03:30
English to Spanish
+ ...
TOPIC STARTER
Stepan Konev wrote: I don't know how to do that with Multiterm, but you can use a plugin for that. Try this video.
IDK if it can help me with the translation memories I have. I mean, can I open TMs as a sdlxliff file?
What I know.is that this tool comes in handy for pretranslation terminology search in order to keep consistency during translation properly.
| | | |
Ivan Chamme
Local time: 03:30
English to Spanish
+ ...
TOPIC STARTER
Hans Lenting wrote: Stepan Konev wrote: I don't know how to do that with Multiterm, but you can use a plugin for that. Try this video. But you’ll have to translate the extracted terms manually. She didn’t even use the concordance… See here for an alternative approach: https://www.star-ts.com/translation/dynamic-linking/ [Edited at 2022-11-16 23:17 GMT]
Just asking, can I download this tool as an isolated app or is it part of another CAT tool?
| | | |
| Pages in topic: [1 2] > |
Login to reply/comment 
Translation news related to SDL Trados
