Studio 2009: Main Translation Memory not updating
Thread poster: philwhite

philwhite
Local time: 23:01
German to English
Feb 27, 2012

I have encountered the following bizarre behavior in studio 2009.

Project set up with two databases. Database 1 is explicitly for the product that I am
translating, database 2 is for other products from the same company, which can be very
similar, but where the terminology used sometimes differs.

Database 1 is set up for Lookup, Concordance, Update and no penalty.
Database 2 is set up for Lookup, Concordance and a penalty of 1 (no Update).

Studio correctly matches segments from Database 2 at 99% if they are identical.

Now comes the interesting part! If I accept a match from database 2 without editing it, it does
not get added to Database 1. Nor do such segments get added if I use the "Update Main
Translation Memory" function. The only way I can get them in where they belong is by
importing the translation files at the end of the job.

The upshot of this is that I do not get 100% matches in future documents, even though the particular segment has been translated and confirmed in the current document. Of course, importing the SDLXLIFF file at the end of the jobs is a workaround, but it is easy to forget.

Also, even if I do import such segments into Database 1, I still do not get 100% matches. Instead, I get a 99% match from Database 1 (multiple translation penalty) and a 98% match from Database 2 (multiple translation penalty and Translation Memory penalty). The same applies if I use attribute filters, namely that the multiple translation penalty is always applied, even if a single given translation matches all my criteria and none of the others do. Surely the purpose of filters and penalties is to allow different translations for different circumstances, and if you filter on those circumstances, no penalty should be applied for multiple translations if only one of the translations fully meets the specified criteria.

I believe that I should be able to switch off the multiple translation penalty, but, frankly, with more than a quarter of a million segments across the two translation memories, and some 30 different products Matthau used as a filter attribute, I am not entirely sure about the resulting behavior.

Thus, there are two issues:
1) Translation Memory 1 does not update if I accept a translation from Translation Memory 2 without editing it.
2) Multiple translation penalties are applied when they are inappropriate.

Has anyone encountered this before? I have searched here and not seen this behavior discussed.

Thanks in advance for any help.


 

Roy OConnor
Local time: 23:01
German to English
Field values may be important Feb 28, 2012

Hi, Phil,

Have you had a look on the SDL Knowledge Base at: http://kb.sdl.com/ under Trados Studio (enter Multiple TMs)? There is some useful info there.

One thing that seems to be important and has caused me quite a lot of trouble is the issue of field values. Multiple TMs should preferably have matching field values, otherwise segments may not be transferred properly. This may or may not be influencing your problem, but it is worth bearig in mind.

Roy


 

philwhite
Local time: 23:01
German to English
TOPIC STARTER
Thanks, but still searching for an answer Feb 28, 2012

Thanks for that. I am, indeed, using fields and filtering on fields, but the two databases have an identical set of fields. There should be no entries in database 2 with the field I am filtering on, but it nevertheless exists. I can really see no explanation for the behavior described above. I shall certainly have a closer look.

Thanks also for the heads-up on the knowledge base. I have to admit that it Has become tiresome trying to wade through the morass of information to find the nugget you need since I lost most of my sight last year!

If I solve the problem, I shall post here.


 

SDL Community  Identity Verified
United Kingdom
Local time: 23:01
English
Interesting Problem Feb 29, 2012

Hi Phil,

I spent 20 minutes or so trying to repro this issue this morning and I can't. Database #1 is updated with the contents of the match from Database #2 as long as I confirm it. So I don't edit it, as you stated, but I still have to confirm it or I am not sending anything to the Translation Memory (You probably did this but I wanted to make sure).

I also tried updating the Main Translation Memory without confirming the hit from Database #2 and this also worked as long as I make sure I have set the option to update draft segments... because of course I didn't confirm it;


Maybe this will help you with your investigation?

Regards

Paul


 

philwhite
Local time: 23:01
German to English
TOPIC STARTER
Curiouser and curiouser Feb 29, 2012

Thank you for your suggestion and for the time you spent. I always confirm all segments as I work through a translation, and this was no exception. When I tried to update the translation memory manually, I even set all the flags, including the draft flag (in the knowledge that the translation was 100% confirmed). I am beginning to suspect after all that there is a problem with the fields, although for the life of me I cannot see what it might be.

There are, of course, numerous workarounds, but that is not the point.

I shall get to the bottom of this eventually!


 

philwhite
Local time: 23:01
German to English
TOPIC STARTER
Solved?? Feb 29, 2012

I just took the database penalty off Database 2, and everything appears to work as expected. I can now confirm an already confirmed fuzzy segment, and it is now added to Database 1.

The filter penalty should be sufficient to guarantee that only segments relating to this product are regarded as 100%. Nevertheless, I was using the "belt and braces" principle.

The introduction of multiple translation memories seemed to me to be an ideal opportunity to thin out some of my very unwieldy databases (several of which contain more than a quarter of a million segments with quite complex field values). This was one such case in point. The product in Database 1 was originally in a larger database which I split to form Database 1 and Database 2.

It seems to me that one should either use multiple databases with a database penalty or a single database with field filters when dealing with related products. Mixing database penalties with filters does not seem to be a good idea.


 


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


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

Studio 2009: Main Translation Memory not updating

Advanced search







TM-Town
Manage your TMs and Terms ... and boost your translation business

Are you ready for something fresh in the industry? TM-Town is a unique new site for you -- the freelance translator -- to store, manage and share translation memories (TMs) and glossaries...and potentially meet new clients on the basis of your prior work.

More info »
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 SDL Trados, MemoQ, Wordfast & major CAT tools. Download and start using CafeTran Espresso -- for free

More info »



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