best way to organize TM
Thread poster: lauradiel
lauradiel
lauradiel  Identity Verified
Australia
Local time: 23:12
Member (2009)
German to Italian
+ ...
Nov 1, 2009

Do you have any suggestions on how to organize TM?

I divide my glossaries into language pairs + subjects, but what about translation memories? Do you create only project specific TM or also bigger ones (one for each client/subject/language pair)?

Thanks


 
Lucas Rayel
Lucas Rayel  Identity Verified
Brazil
Local time: 10:12
English to Portuguese
+ ...
Huge TM Nov 1, 2009

Hello,

I just have one huge TM with everything I've ever translated... As I translate a lot of projects and documents for university students and they don't even know what CAT tools are, they don't care about the TM aswell.
With that said, I can manage to have all in one TM. But if someday I get a real company client, then if it specifies that they want the TM from their texts, then I'd do it... If not, I'd just put all together with my existent TM.

Lucas


 
Kevin Lossner
Kevin Lossner  Identity Verified
Portugal
Local time: 13:12
German to English
+ ...
Depends on your tools Nov 1, 2009

The best strategy for organizing your TM depends in a large part on what CAT tool you use. Some have performance limitations as the size of the TM grows, and only a few are able to differentiate content for matching according to a hierarchy or client choice.

 
Jan Willem van Dormolen (X)
Jan Willem van Dormolen (X)  Identity Verified
Netherlands
Local time: 14:12
English to Dutch
+ ...
Double strategy Nov 1, 2009

I use a TM for each company I work for, plus I put everything after I'm done with a project into one huge TM. In Trados, I've set (under Options > Translation Memory Options > Concordance) this big TM as reference TM, so that, whenever there's no match in the companyspecific TM I'm using, Trados looks in the background TM. That way, matches that are not companyspecific turn up too.
Also, with the company TM's, I always set a text field with that company's name (Settings > Project and Filte
... See more
I use a TM for each company I work for, plus I put everything after I'm done with a project into one huge TM. In Trados, I've set (under Options > Translation Memory Options > Concordance) this big TM as reference TM, so that, whenever there's no match in the companyspecific TM I'm using, Trados looks in the background TM. That way, matches that are not companyspecific turn up too.
Also, with the company TM's, I always set a text field with that company's name (Settings > Project and Filter Settings), so in the large background TM, Trados can distinguish between translations for different clients.
Collapse


 
Amal Ibrahim
Amal Ibrahim  Identity Verified
Germany
English to Arabic
Performance limitations.. Nov 1, 2009

Kevin Lossner wrote:
Some have performance limitations as the size of the TM grows, and only a few are able to differentiate content for matching according to a hierarchy or client choice.


Hi Kevin,

Would you elaborate?

In your opinion, what would be the optimal TM size for MemoQ and SDL Trados/Studio? I guess DVX does not have this kind of limitation.

Thank you so much!


 
Samuel Murray
Samuel Murray  Identity Verified
Netherlands
Local time: 14:12
Member (2006)
English to Afrikaans
+ ...
It evolves, I think Nov 2, 2009

Laura Diel wrote:
Do you have any suggestions on how to organize TM?


It evolves, I think. It also depends on your TM tool and your client base and the types of documents you commonly translate. What works for me may not work for you. I have a new TM for each new client and each new job, unless it is for a client who often sends the same type of file, in which case I have a TM for all of that client's work, or unless I have several clients who send the same type of file, in which case I have a TM for the subject field. Also, when I get a new client in a particular field, I may do a search on my old TMs to see if any of them are likely to help me, and then I merge them and penalise them for maximum benefit.


 
lauradiel
lauradiel  Identity Verified
Australia
Local time: 23:12
Member (2009)
German to Italian
+ ...
TOPIC STARTER
Wordfast Nov 2, 2009

I normally use Wordfast. As far as I know, you can select only one TM at a time with this CAT tool (I will check in the manual). Maybe it would be a good idea to have one huge TM for each language pair and then smaller ones for each company/agency. Are there any performance limitations with Wordfast? And doesn't the TM become too big with the time?

 
Sergei Leshchinsky
Sergei Leshchinsky  Identity Verified
Ukraine
Local time: 15:12
Member (2008)
English to Russian
+ ...
custom TMs Nov 2, 2009

Laura Diel wrote:
Do you create only project specific TM or also bigger ones (one for each client/subject/language pair)?

I have big TMs for big running projects and small field-specific TMs. All stored in TMX (except for those in use). Plus, I have several cumulative TMs on several broad areas like Software Localization, Technologies... They are use as second (supplementary) read-only TMs for concordance search. (like Jan Willem van Dormolen said above)

If I get a TM along with the order and the files to be translated, I use that TM as a basis for the temporary working TM and import several related TMs from my archive into it to make it bigger. This combined TM is a temporary one. (The original TM received from the client is kept ZIPped.) After the job is done, the TM deleted and the project is cleaned into one of the cumulative TMs. The original TM is sometimes imported there to.

So, the sizes of TMs are increasing with each job. TM management is very important.
Some tasks (like killing repetitions) can be solved using Apsic Xbench.

Any other hints?

[Редактировалось 2009-11-02 16:17 GMT]


 
Tomás Cano Binder, BA, CT
Tomás Cano Binder, BA, CT  Identity Verified
Spain
Local time: 14:12
Member (2005)
English to Spanish
+ ...
Customer and end customer Nov 2, 2009

Something you really not want is the name of a wrong company or product in the materials you deliver to their real or potential competitors. You certainly don't want wrong terminology either if your customers ask you to use different sets of terms or glossaries. So using one single TM is not an option in my opinion.

So to me, independently of the tool, the best you can do is to have a memory for each of the customer+end customer combination.

For terminology, you can use
... See more
Something you really not want is the name of a wrong company or product in the materials you deliver to their real or potential competitors. You certainly don't want wrong terminology either if your customers ask you to use different sets of terms or glossaries. So using one single TM is not an option in my opinion.

So to me, independently of the tool, the best you can do is to have a memory for each of the customer+end customer combination.

For terminology, you can use glossaries or termbases. But for actual text you certainly want to keep firm boundaries between customers (and their end customers if you work for agencies).
Collapse


 
Kevin Lossner
Kevin Lossner  Identity Verified
Portugal
Local time: 13:12
German to English
+ ...
Depends on how you want to work Nov 2, 2009

Amal Ibrahim wrote:
Hi Kevin,

Would you elaborate?

In your opinion, what would be the optimal TM size for MemoQ and SDL Trados/Studio? I guess DVX does not have this kind of limitation.


DVX really doesn't have much limitation in this regard; I've been able to use large TMs (500,000+ units) well with that tool. The problem is moving large chunks of data in and out for maintenance work with other tools. Even though I used to program large stored procedures in SQL and use the language for other purposes, with DVX it drives me nuts, though the excellent guide some fine volunteer wrote usually gets me through. (I forget what it's called - it's available for download on a Canadian site I think.)

I also haven't seen performance problems with Trados - there the problem is selecting the right match for the customer, and one is pretty much forced to use the customer-specific TM with a big reference in the background. The problem comes if you have more than one big reference to use. On the plus side, import/export from TWB has always been reasonably fast, unlike DV.

In general, I prefer MemoQ for my work these days, though I find I am hindered by the inability to export my project comments. (This will be addressed in v 4.x.) However, when I migrated my big DVX TM to MemoQ I was shocked at how slow the tool ran. Even opening projects attached to that TM takes a long time. That really bugs me. The developers tell me that the database is "optimized" for a smaller TM, but I bloody well want to re-optimize it for my way of working. Import/export times fall between the never/forever rate of DVX for large data quantities and the zippy performance of TWB - acceptable to me. With smaller TMs I've used (tens of thousands of units), MemoQ is great. The delays I'm griping about here are not huge, but I notice them with easy texts, and I find it upsetting to wait for a minute or more when I want to open a project just to check some little thing. On the major plus side, MQ doesn't limit the number of TMs you can use - it's like the DVX Workgroup version in many ways, but at a quarter the price. And it's the only one of the three I mentioned with reasonable support. Atril seems dead in that regard (despite the PowerLing partnership - real support includes development), and SDL Trados is... well, SDL Trados. It's victims... uh, users... know what that means. There are occasional bright lights in that firmament - I think the support guy assigned to cover the ProZ forum is pretty good - but even agency customers with support contracts have complained bitterly to me in private about the lack of response or useful response. My experiences are mixed - some awful, some great, but still not even close to Kilgray these days.


 
Jamie Lucero
Jamie Lucero  Identity Verified
United States
Local time: 06:12
French to English
TMs and performance in Wordfast Nov 3, 2009

Laura Diel wrote:

I normally use Wordfast. As far as I know, you can select only one TM at a time with this CAT tool (I will check in the manual)...Are there any performance limitations with Wordfast?


Wordfast Classic (MS Word application) allows you to use up to 3 TMs simultaneously: a primary local TM, a background or reference local TM, and a remote or server TM (AKA VLTM). Wordfast Pro (standalone Java application) allows you to use an unlimited number of local and remote TMs, and they can all be used as primary or reference TMs. In addition, the TMs can be put in any order of preference.

Because Wordfast TMs are just tab-delimited text files and store minimal formatting information, performance tends to remain quite high despite really large TM sizes. This, of course, is limited to your own system's performance, and in the case of Classic, the memory load in MS Word.

As a quick performance test, I just imported an old 400,000+ TU (400+ MB) Trados TMX file into Pro in less than 5 minutes. There was no noticeable speed decline in pretranslating a 7-page word document, which took about 1 minute.

Jamie


 
Joel Earnest
Joel Earnest
Local time: 14:12
Swedish to English
Mulitple TMs Nov 3, 2009

With Studio, I normally run my main TM along with any customer-specific TM I might have, setting a five percent penalty on the main TM so that hits in the customer-specific TM will always be prioritized.

 
Pablo Bouvier
Pablo Bouvier  Identity Verified
Local time: 14:12
German to Spanish
+ ...
Best way to organize TM Nov 3, 2009

Tomás Cano Binder, CT wrote:

Something you really not want is the name of a wrong company or product in the materials you deliver to their real or potential competitors. You certainly don't want wrong terminology either if your customers ask you to use different sets of terms or glossaries. So using one single TM is not an option in my opinion.

So to me, independently of the tool, the best you can do is to have a memory for each of the customer+end customer combination.

For terminology, you can use glossaries or termbases. But for actual text you certainly want to keep firm boundaries between customers (and their end customers if you work for agencies).


IMHO, all options are correct according to his utility.
But, I believe we lose easily of sight the utility of translation memory systems.

If I write "open the door of the vehicle", I will be able to use this translation for "any vehicle". It does not matter if the vehicle is a car, a plane, or a rockeIf. But, if I write open the door of your [mark, brand, etc], this sentence will lack of value in other contexts and I will limit myself to a brand or at least to a certain type of vehicles (for example, cars)...

Personally, I believe we should work as possible with a neutral (unpersonalised) reference memory.


 
lauradiel
lauradiel  Identity Verified
Australia
Local time: 23:12
Member (2009)
German to Italian
+ ...
TOPIC STARTER
thank you Nov 3, 2009

Thank you for all the useful suggestions, there are several aspects to take into account and I need to see what works best for me. Thanks Jamie for the clear explanation!

 
Karine Yacoubian
Karine Yacoubian  Identity Verified
Cyprus
Local time: 15:12
Greek to English
+ ...
Penalising TMs when merging and other queries Mar 12, 2018

Samuel Murray wrote:

Laura Diel wrote:
Do you have any suggestions on how to organize TM?


It evolves, I think. It also depends on your TM tool and your client base and the types of documents you commonly translate. What works for me may not work for you. I have a new TM for each new client and each new job, unless it is for a client who often sends the same type of file, in which case I have a TM for all of that client's work, or unless I have several clients who send the same type of file, in which case I have a TM for the subject field. Also, when I get a new client in a particular field, I may do a search on my old TMs to see if any of them are likely to help me, and then I merge them and penalise them for maximum benefit.



Hi Samuel,

This may have been several years ago but if you (or anyone else, for that matter) have a moment to explain, I'd be very grateful if someone could clarify what it means when one penalises TMs for maximum benefit - benefit of what? And how do you penalise them? How does that work and how does it help me? I am only aware of penalisation when aligning TMs.

I work with tools like Wordfast, MemSource and others, but am currently setting up TMs and TBs for a newly purchased SDL Trados 2017 licence. I've got a lot of source and target texts to align and have to organise TMs, and I have to say that your take on this has been extremely helpful. As you have shown in your post, it all depends on the needs of your clients and what type of documents you get most often at the end of the day.
With Trados, I wasn't sure how to organise it all (by client, subject, etc) and worried about confidentiality too. Any advice on that? I guess that keeping different client TMs separate is a good idea not just in terms of organisation and professionalism, but also for confidentiality and data protection, and my responsibility to uphold that agreement as a translator!

Any help answering my queries would be greatly appreciated.

Karine


 


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


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

best way to organize TM







Anycount & Translation Office 3000
Translation Office 3000

Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.

More info »
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 »