Pages in topic:   [1 2 3 4] >
How do you organise translation memories?
Thread poster: Dan Lucas

Dan Lucas  Identity Verified
United Kingdom
Local time: 07:31
Member (2014)
Japanese to English
Sep 25, 2014

I'm considering various ways of slicing and dicing my translation memories. One approach would be to have just one TM for everything, a sort of MyBigFatTM.sdltm. At the other end of the scale one might have a number of finely divided TMs, such as Electronics.sdltm, PatentsElectronic.sdltm, ManualsElectronics.sdltm and so on.

Currently I'm leaning towards having one large TM. Are there any disadvantages to this approach? How do other translators go about this?

For what it's worth, I'm using SDL Trados Studio 2014, but this is not a question specific to any particular CAT package.

Thanks
Dan


Direct link Reply with quote
 

Catarina Lopes
Portugal
Local time: 07:31
Member (2013)
English to Portuguese
+ ...
Thank you, Dan, for your question. Sep 25, 2014

I'm a new Trados user and I have also been wondering what's the best way to organise my TMs.

Like you, I'm also considering creating one big TM per language pair. I have read you can add attributes in order to filter searches, but I'm not yet sure how to do this.

Let's wait and see what our colleagues have to say about this.

Best regards,

Catarina Lopes


Direct link Reply with quote
 

Denise Phelps  Identity Verified
Local time: 08:31
Spanish to English
+ ...
BigFatTM Sep 25, 2014

I also use Trados Studio 2014 (and previously 2011), and went with the BigFatTM approach, because I'm never given or asked to hand in project TMs (though I could always set those up separately, if the need ever arose) and I don't get much repetition in my work so most of my leverage comes from concordance, multi term and QA. So for me it makes sense to have everything in the same place rather than adding a zillion TMs to each project. It also helped me reach the x amount of segments you need to generate an autosuggest dictionary (which would be more useful if I could type without looking at the keyboard). It also saves time when setting up projects, and headaches when deciding how to classify everything in a way that will still make sense to me 12 months later.

But I asked this same question some time ago with no response, so I'll be interested to see if other people answer with other ideas.


Direct link Reply with quote
 

Samuel Murray  Identity Verified
Netherlands
Local time: 08:31
Member (2006)
English to Afrikaans
+ ...
Depends on your CAT tool Sep 25, 2014

Dan Lucas wrote:
I'm considering various ways of slicing and dicing my translation memories. One approach would be to have just one TM for everything, a sort of MyBigFatTM.sdltm. At the other end of the scale one might have a number of finely divided TMs, such as Electronics.sdltm, PatentsElectronic.sdltm, ManualsElectronics.sdltm and so on.


Having separate memories and separate glossaries and separate reference materials for separate clients or separate subject fields is certainly a useful thing, but doing it depends on how easy it is in your chosen CAT tool to link each collection of TMs etc to each new job that comes in. I'm afraid I don't know exactly how Trados 2009/11/14 deals with this, so I can't advise you on that.

I use WFC, and WFC is a project-less CAT tool. Instead of projects, it has settings files. You can set up the program to use a certain TM, certain glossaries, etc, and then save that configuration as a settings file. Then, when a new job comes in, you select which settings file you want applied to it, and instantly all the TMs, glossaries etc are associated with that new job. If I know that a certain client has specific requirements (terminology etc) then I create a separate settings file for it. I have a general settings file for pharmaceutical work, for example, but I also have a number of pharmaceutical clients who have very specific requirements, and for them I have specially set up settings files. Of course, it's possible to use a single TM in multiple settings, so even though I use a client-specific TM for a specific client, I can still benefit from the general TM in that field as well.

And, to make sure I don't accidentally break my settings files, I store master copies of them elsewhere and let a batch file copy them into the WFC folder (overwriting any old, broken ones) every time I reboot.


Direct link Reply with quote
 

Dan Lucas  Identity Verified
United Kingdom
Local time: 07:31
Member (2014)
Japanese to English
TOPIC STARTER
Ways to export part of a TM? Sep 25, 2014

I also use Trados Studio 2014 (and previously 2011), and went with the BigFatTM approach, because I'm never given or asked to hand in project TMs (though I could always set those up separately, if the need ever arose)

Thank you for your comments. I guess in Studio you could use filters to export a subset of the main TM if a client wanted it?

Dan


Direct link Reply with quote
 

Denise Phelps  Identity Verified
Local time: 08:31
Spanish to English
+ ...
Filters Sep 25, 2014

Filters? I'm not so high tech

If a client wanted a TM at the end of the project, in principle I could create a new one at the start, running concurrently with my BigFat TM. Actually, no client has ever requested this. But if they did, I think you create a new TM when you set up the translation, make sure it's ticked for updated, and go ahead. You leverage your BigFat TM and feed into your client specific TM.

[Edited at 2014-09-25 22:08 GMT]


Direct link Reply with quote
 

Samuel Murray  Identity Verified
Netherlands
Local time: 08:31
Member (2006)
English to Afrikaans
+ ...
@Denise Sep 26, 2014

Denise Phelps wrote:
You leverage your BigFat TM and feed into your client specific TM.


How does Trados do it... if a segment in the BigMomma TM is a 100% match and you don't edit the segment (just accept it), does Trados then add that segment to the client-specific TM?


Direct link Reply with quote
 

Natalia Kobzareva  Identity Verified
Russian Federation
Local time: 09:31
Member (2009)
English to Russian
+ ...
I combine two approaches Sep 26, 2014

I have my Reference TM accumulated during my first years as a Trados user when I still had no solution how to organise.
This TM is used for concordance only and is not updated.
I have also four major TMs: Court, Securities, Agreements, Corporate, and a number of smaller ones created to produce some specific translations.
With Multiterm I create a new termbase for each project/document (that may contain from 3 to 20 pages of terms) for consistency purposes. After the document is done, I add the entries of this specific termbase into my Reference Termbase, delete duplicates (if any), and delete this particular termbase.


Direct link Reply with quote
 

Denise Phelps  Identity Verified
Local time: 08:31
Spanish to English
+ ...
@ Samuel Sep 26, 2014

No idea: I've never had to do it. But I imagine I would have to accept/reconfirm the segment to make sure it was entered into the new TM.

Direct link Reply with quote
 

Dan Lucas  Identity Verified
United Kingdom
Local time: 07:31
Member (2014)
Japanese to English
TOPIC STARTER
That's an interesting way of doing things Sep 26, 2014

Natalia Kobzareva wrote:
I have my Reference TM accumulated during my first years as a Trados user when I still had no solution how to organise.
This TM is used for concordance only and is not updated.
I have also four major TMs: Court, Securities, Agreements, Corporate, and a number of smaller ones created to produce some specific translations.
With Multiterm I create a new termbase for each project/document (that may contain from 3 to 20 pages of terms) for consistency purposes. After the document is done, I add the entries of this specific termbase into my Reference Termbase, delete duplicates (if any), and delete this particular termbase.

So you have a Reference TM and a Reference Termbase and you update the Termbase but not the TM?

How about your four major TMs, I guess these are updated regularly as you go through projects?

Dan


Direct link Reply with quote
 

Giles Watson  Identity Verified
Italy
Local time: 08:31
Italian to English
Pick 'n' mix Sep 26, 2014

Samuel Murray wrote:

Having separate memories and separate glossaries and separate reference materials for separate clients or separate subject fields is certainly a useful thing, but doing it depends on how easy it is in your chosen CAT tool to link each collection of TMs etc to each new job that comes in. I'm afraid I don't know exactly how Trados 2009/11/14 deals with this, so I can't advise you on that.



I use Studio 2014.

Most of my work is organised into client-specific projects, to which I add new jobs as they crop up.

This approach enables me to add or detach termbases and translation memories as necessary, ticking the update box if appropriate when I attach the TM. For example, if a finance-oriented job comes in from a wine client I might add one or two extra TBs/TMs (for concordance purposes) just for that document and remove them for the next, more wine-specific, text to reduce "noise" and speed up search times.

One approach that seems to be useful with a lot of jobs is to attach the appropriate IATE glossary, - thanks to Paul Filkin! - as a MultiTerm termbase AND as a Studio translation memory. You can convert IATE's tbx to tmx for import into a Studio TM using the Glossary Converter app.


Direct link Reply with quote
 

Carole Paquis  Identity Verified
United Kingdom
Local time: 07:31
Member (2007)
English to French
Mixed economy too Sep 26, 2014

I work with Trados Studio 2011.

I also have a mixed economy approach: 1 tm + 1 termbase per client, to which is also added a BigFat tm. I find this is flexible and efficient. It maximises the probability of concordance. That way, I can use client specific phrases but also leverage good ideas I have had for other clients...
For each new project, I end up with 2 or 3 TMs working in the background.

The other advantage of also having a BigFat tm is that it is big enough to have an Autosuggest dictionary.


Carole PAQUIS


Direct link Reply with quote
 

Dan Lucas  Identity Verified
United Kingdom
Local time: 07:31
Member (2014)
Japanese to English
TOPIC STARTER
All useful feedback Sep 26, 2014

Thank you Giles, Carole and of course Samuel, this is very useful.

If anybody else would like to comment we'd love to hear how you do it!


Direct link Reply with quote
 

Natalia Kobzareva  Identity Verified
Russian Federation
Local time: 09:31
Member (2009)
English to Russian
+ ...
Yes, Dan Sep 26, 2014

I update my major 4 TMs regularly while doing translations.
Reference Termbase gets updated quite seldom now, as almost all common legal terms are already there.


Direct link Reply with quote
 

Michael Joseph Wdowiak Beijer  Identity Verified
United Kingdom
Local time: 07:31
Member (2009)
Dutch to English
+ ...
Sub-segment based auto-assembly Sep 26, 2014

For concordance: I have a massive TM database in TMLookup, comprising 50,000,000 TUs. This includes András Farkas’ amazing EU TM collection (CELEX collection, EP collection, Press release collection). Info: http://www.farkastranslations.com/eu_translation_memories.php + lots of stuff from the OPUS site: http://opus.lingfil.uu.se/

My TMX folder: I also keep all my TMs from the last few years in a folder structure on my computer, all organised into clients, end clients, etc. I can create special TMs for specific jobs by combining (parts of) these using Okapi Olifant, the Heartsome TMX editor, the built in TMX editor in CafeTran and/or a text editor.

Pretranslation in CafeTran: I usually pretranslate each new job against one of my Big Mama TMs (actually called ‘Big_Monkey.tmx’) in CafeTran before starting a job, to see if I happen to have any useful matches in any of my TMs. CafeTran has a feature that allows you to run a pretranslation and then export the results to a TMX file. This means you can then detach the large TMX and reattach the much smaller TMX with only the matches. CT calls these ‘fragments’ or ‘subsegments’. They are referred to as LSCs (longest substring concordance) in memoQ, and sth else (can't remember) in DVX.

Project TM: Each job also gets its own Project TM.

Termbases: However, the real magic happens in TBs, rather than TMs. I tend to focus much more on my termbases than on my TMs. I keep changing my TB workflow, but currently work with three main ones:

1. Glossary.csv (around 400,000 entries; in need of some serious cleaning)
2. stop words.csv (the, and, to, etc.; hits from Glossary.csv get highlighted in the src box, whereas stop word hits don't: this makes it easier to focus on useful hits)
3. tmp.csv (this contains several very large TBs: the entire recent IATE database (thanks Henk Sanderson!!!), the Apple glossary, EU legislation glossary (http://www.farkastranslations.com/glossaries.php ), etc.; in CafeTran you can switch ON/OFF matching for a TB on the fly via the right-click ‘Stop automatic matching’ switch)

I am also constantly extracting term pairs from TMs, both from my own TMs and from reliable TMs I find here and there (sometimes with SynchroTerm). Some of the DGT stuff, the CELEX stuff, etc.

The general idea is to move from relying on TMs to using only TBs, as they offer much more fine-grained control when it comes to auto-assembly.

Several CafeTran users have already managed to achieve amazing results using a combination of TBs and CafeTran's auto-assembly system. Personally, I still find I get better results from Google Translate, but that is probably due to two things: (1) my main termbase is a real mess, and (2) I translate too many different kinds of texts. If you translate mainly one kind of text, and you put some work into creating a good TB, CafeTran can auto-assemble your translation up to 75% or even more correctly. Sort of a home-made MT system: ‘Sub-segment based auto-assembly’ or ‘sub-segment based machine translation’.

Another interesting development is that CafeTran has a new feature that tries to repair your auto-assembly results using … machine translation (currently Google Translate, Microsoft Translator, or My Memory). Yes, kind of like that thing that DVX (and no one else yet) can do.

Michael

PS: Daniel Benito (aka Mr DVX/Atril) wrote a very good article about some of these things: http://www.fti.uab.cat/tradumatica/revista/num7/articles/07/07central.htm


Direct link Reply with quote
 
Pages in topic:   [1 2 3 4] >


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


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

How do you organise translation memories?

Advanced search







SDL MultiTerm 2017
Guarantee a unified, consistent and high-quality translation with terminology software by the industry leaders.

SDL MultiTerm 2017 allows translators to create one central location to store and manage multilingual terminology, and with SDL MultiTerm Extract 2017 you can automatically create term lists from your existing documentation to save time.

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 »



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