Pages in topic:   [1 2] >
Trados 2009 - no 100% match for already translated segments
Thread poster: Radian Yazynin

Radian Yazynin  Identity Verified
Local time: 13:39
Member (2004)
English to Russian
+ ...
May 4, 2011

Hi colleagues,
I need help to see what may happen in this situation. I have created a project and translated all multiple ttx files in it. Then I updated the main TM. After that I tried to pre-translate the same batch of files (identical files before processing in Trados) I didn't recieve a 100% result, i.e. some segments fell out and I had to manually complete the task.
In particular, if I select a segment, translate it and confirm it (!) it should be in the TM, but when I choose an
... See more
Hi colleagues,
I need help to see what may happen in this situation. I have created a project and translated all multiple ttx files in it. Then I updated the main TM. After that I tried to pre-translate the same batch of files (identical files before processing in Trados) I didn't recieve a 100% result, i.e. some segments fell out and I had to manually complete the task.
In particular, if I select a segment, translate it and confirm it (!) it should be in the TM, but when I choose an absolutely the same segment in a different file I first get a - say - 98 or 99 % match that I have to accept and confirm. Where is that 100% match I have already confirmed then? I have also tried to set a penalty - no way.
The main TM was properly updated.
Thanks.
Collapse


 

Kathleen Misson  Identity Verified
Spain
Local time: 12:39
Member (2009)
Spanish to English
I have the same problem May 9, 2011

Hope someone can post an answer. Many thanks.

 

Ulf Samuelsson  Identity Verified
Mexico
Local time: 05:39
English to Swedish
+ ...
Same problem May 9, 2011

I had the same problem several times on Friday.
The memory had 78,000 segments, and I often got 85% matches or 90% matches, but when searching the memory I found 100% matches for the complete segment that hadn't been proposed. Sometimes I even got no match at all, but when searching the memory I still found a 100% match for the segment.
Only one memory activated, and only one Word file being translated.


 

SDL Community  Identity Verified
United Kingdom
Local time: 12:39
Member (1970)
English
Can we see more examples? May 9, 2011

Hi,

Can you share a screenshot with all tags displayed showing the segment in the Studio Editor and the Translation Memory window with the failed match? Ideally I would have a copy of your Project complete with TM so I could see first hand, but this might shed some light in the first instance.

Regards

Paul


 

Pavel Tsvetkov  Identity Verified
Bulgaria
Local time: 13:39
Member (2008)
English to Bulgarian
+ ...

MODERATOR
Same here... May 9, 2011

... and this has been reported many times on this forum. It is often the case that I know a segment has been translated, but Trados simply cannot find it until I do a manual search.

 

SDL Community  Identity Verified
United Kingdom
Local time: 12:39
Member (1970)
English
Can I see it? May 10, 2011

Hi Pavel,

This may have been reported on the forum but I think I'm correct in thinking we (SDL) have not been given anything to help us verify this behaviour without a reasonable explanation for why the problem is thought to exist.

If anyone can give me a document and a TM and instructions on what to do so that I can experience the same thing I would be very happy to get the problem addressed.

Regards

Paul


 

Radian Yazynin  Identity Verified
Local time: 13:39
Member (2004)
English to Russian
+ ...
TOPIC STARTER
Answer to Paul May 11, 2011

Hi Paul, thanks for your reply and sorry for the delayed feedback (anyway I am the thread poster). You are right - we should have been more attentive to the smart tag replacement feature, and maybe something was ignored in my case after which I didn't get that anticipated result. Maybe the TM (over 150K words) was corrupted in some segments (due to abnormal termination of Trados in some cases). At least the newly created has shown all the 100% matches I wanted to find. This is my situation and i... See more
Hi Paul, thanks for your reply and sorry for the delayed feedback (anyway I am the thread poster). You are right - we should have been more attentive to the smart tag replacement feature, and maybe something was ignored in my case after which I didn't get that anticipated result. Maybe the TM (over 150K words) was corrupted in some segments (due to abnormal termination of Trados in some cases). At least the newly created has shown all the 100% matches I wanted to find. This is my situation and if somebody can say more I would still be very much grateful for that.
Using this opportunity, I would also like to kindly ask Paul about one more issue.

My background:

Windows 7 Pro, Russian version.
Latest Trados 2009 update installed.

This topic has been already discussed in some other thread, I know. But this is also a problem with 100% matches not being used properly. I still can't get rid of number mismatches in the pair EN-RU or DE-RU.
For example, I have many segments containing "Code x.xx" and my TM does offer a substitution of numbers in the Concordance window (deleting the old numbers after "Code" and offering the contextually correct numbers) while in the target segment - not!
Also simply copying the source segment to target (preserving the original numbers) and translating only the textual part results in a message like "missing numbers" or "extra numbers" in target segment.
This prevents me from proper verification and distracts very much.
Could you suggest something as to how avoid such situation?
Thanks so much.
Collapse


 

Jüri Liivak  Identity Verified
Estonia
Local time: 13:39
Member (2010)
English to Estonian
+ ...
Almost the same problem May 12, 2011

Hello,
I have almost the same problem. Trados 2009 editor didn't show all 100% matches, but if I search this sentence manually for concordance, the translation appears in concordance window as 100% match. I sticked off all penalties and generated this project again, but problem remains (this sentence didn't apper in Translation results window). This window also didn't show results with less than 100% sentences (99-70%). It seems to me that it is some kind of Trados database search problem
... See more
Hello,
I have almost the same problem. Trados 2009 editor didn't show all 100% matches, but if I search this sentence manually for concordance, the translation appears in concordance window as 100% match. I sticked off all penalties and generated this project again, but problem remains (this sentence didn't apper in Translation results window). This window also didn't show results with less than 100% sentences (99-70%). It seems to me that it is some kind of Trados database search problem.
Nice to hear opinions.
Collapse


 

Pavel Tsvetkov  Identity Verified
Bulgaria
Local time: 13:39
Member (2008)
English to Bulgarian
+ ...

MODERATOR
Paul has always been great help May 13, 2011

Paul from SDL has helped me out on several occasions when help was urgently needed, so I do not want to get into any arguments with him, but I have to say that I finished work on a translation project just yesterday and this happened to me more than once. My percentage limit for filling in previous translations is set to 75% and nevertheless 95%+ or even 100% do not get recognized if they are not looked for manually. This does not... See more
Paul from SDL has helped me out on several occasions when help was urgently needed, so I do not want to get into any arguments with him, but I have to say that I finished work on a translation project just yesterday and this happened to me more than once. My percentage limit for filling in previous translations is set to 75% and nevertheless 95%+ or even 100% do not get recognized if they are not looked for manually. This does not happen every time, but it is a problem, and, yes, it has been around for quite a long time - and it has not been fixed.

Just by looking at this thread here I cannot help but feel that everyone is aware of the problem, but SDL where it seems that the prevailing feeling still is: the problem is thought to exist.
Collapse


 

RWSTranslation
Germany
Local time: 12:39
Member (2007)
German to English
+ ...
Update Main TM sometimes fails May 13, 2011

I faced such a problem (Segment already translated, but not found) a couple of time too.
As I checked inside the TM, I noticed that the new segments were not added at all in the TM. The batch task "Update Main TM" had not done its job. I had to repeat the task again to get the translation in the TM.

Seems like it is something else for many of you, but this can be an explanation for the others...


 

SDL Community  Identity Verified
United Kingdom
Local time: 12:39
Member (1970)
English
Couple of responses in one ;-) May 13, 2011

@Jüri
It's worth noting that a concordance 100% match is not the same as a Translation 100% match, so there could be various reasons for the discrepancy.

If you search for a single word in concordance which is in the TM, you expect and get a 100% match. However, you usually don't get a 100% match for that search in normal search if the match segment contains additional words compared to the query. That illustrates the key difference in the scoring methods: concordance scori
... See more
@Jüri
It's worth noting that a concordance 100% match is not the same as a Translation 100% match, so there could be various reasons for the discrepancy.

If you search for a single word in concordance which is in the TM, you expect and get a 100% match. However, you usually don't get a 100% match for that search in normal search if the match segment contains additional words compared to the query. That illustrates the key difference in the scoring methods: concordance scoring is based on the words which are common between the match and the query, while normal scoring evaluates all words in the match against all words in the query (i.e. swapped, moved, missing, or inserted words will reduce the score). In concordance scoring, if the match contains all words which are in the query in the same order, you usually get a 100% match. Any words which are in the match but not in the query will not be penalized. You don't expect this behavior in normal search.

If you want to share your results with me I'd be happy to try and explain why?


@Pavel
Me neither I would be happy to investigate this with you though, so if you can share your Project so I have your settings, and give me some steps to repro what you experience, we can look together?


@Radian
My guess is this has something to do with you using different number formats to those that we consider the 'correct' numbers based on this site:

http://msdn.microsoft.com/en-us/goglobal/bb896001.aspx

You can change your O/S to get the correct version and then the language cultures listed will reflect the ones we use. For example taking EN-RU and DE-RU, and I'll pick a single sub-language for this example, I see this for a simple positive number format:

en(US) - 12,346.00
ru(RU) - 12 346,00
de(DE) - 12.346,00

So if you were to be going from en(US) to ru(RU) and the number wasn't recognised, or you entered one that did not match this format then you would see a message telling you the number was missing or incorrectly translated as you describe.

Maybe this is a good place to start?

Regards

Paul
Collapse


 

Radian Yazynin  Identity Verified
Local time: 13:39
Member (2004)
English to Russian
+ ...
TOPIC STARTER
Number formats May 13, 2011

SDL Support wrote:

@Radian
My guess is this has something to do with you using different number formats

Hi Paul, sorry, I am still confused. My default language is Russian and there is precisely the same number format as you stated (eg., ru(RU) - 12 346,00). Standard settings both for Russian and Br. EN for my PC. Could you suggest something? And the problem is not in 'incorrect' reproduction of formats, just simply copying to target (from source) a number like 3.1 results in the error message. Even numbers like '303 848 484' etc. are perceived as missing or extra if even simply copied from source...


 

SDL Community  Identity Verified
United Kingdom
Local time: 12:39
Member (1970)
English
Like this? May 13, 2011

Hi Radian,

You mention copying source to target gets you an error, so let me try and explain what I meant using this example. This is an ru(RU) - en(UK) translation. If I copy source to target and confirm, I get exactly what you describe because 12 346,2 is not a recognised number format for en(UK):


If I just look at the automated localisation of the number
... See more
Hi Radian,

You mention copying source to target gets you an error, so let me try and explain what I meant using this example. This is an ru(RU) - en(UK) translation. If I copy source to target and confirm, I get exactly what you describe because 12 346,2 is not a recognised number format for en(UK):


If I just look at the automated localisation of the number that Studio applies based on the use of an en(UK) target language I see this (using ctrl+, to invoke the Quickplace):


If I now correct the translation and confirm it I see this:


I hope this explains what I meant a little better. It might not be the explanation for your problem, but this is what I meant. Does this help?

Regards

Paul
Collapse


 

Radian Yazynin  Identity Verified
Local time: 13:39
Member (2004)
English to Russian
+ ...
TOPIC STARTER
Copying segments May 13, 2011

[quote]SDL Support wrote:
You mention copying source to target gets you an error/quote]
So, in order to avoid this I should disable the localization feature?
What would you recommend to get a perfect match in a situation:
Source: 'Code 3.1'
Target: (in Russian, preserving the *.* format, i.e. with a decimal point) 'Kod 3.1'
The concordance window offers me translation of 'Code', crosses out the old number and suggests a replacement (the actual number), but Trados doesn't do it automatically and inserts the old numbers in the target ...
Hope I have explained it clearly, if you need screenshots I would prepare them.
Thanks


 

SDL Community  Identity Verified
United Kingdom
Local time: 12:39
Member (1970)
English
Other options May 13, 2011

Hi Radian,

ok - I see what you are trying to do now. I hadn't appreciated they weren't really numbers until now.

Your options, to preserve the autolocalisation aspect, are these:

1. Allow the 3.1 to become 3,1 and then search and replace the entire document afterwards. You might want to disable auto-propogation for 100% matches to avoid incorrect translations in the TM

2. Use a Temporary TM with different language pairs to ensure that 3.1 sta
... See more
Hi Radian,

ok - I see what you are trying to do now. I hadn't appreciated they weren't really numbers until now.

Your options, to preserve the autolocalisation aspect, are these:

1. Allow the 3.1 to become 3,1 and then search and replace the entire document afterwards. You might want to disable auto-propogation for 100% matches to avoid incorrect translations in the TM

2. Use a Temporary TM with different language pairs to ensure that 3.1 stays 3.1.

You can also turn off the QA verification in Tools - Options - verification - QA Checker - Numbers if you want to just avoid the errors being flagged all the time.

Not an elegant solution, particularly if you have lots of other real numbers in the translation. Maybe some of the users here have a better approach as daily practitioners, but these would be my suggestions at the moment.

Regards

Paul
Collapse


 
Pages in topic:   [1 2] >


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

Trados 2009 - no 100% match for already translated segments

Advanced search







Protemos translation business management system
Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!

The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.

More info »
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

More info »



Forums
  • All of ProZ.com
  • Term search
  • Jobs
  • Forums
  • Multiple search