Mobile menu

Translation Memory - how to make the best of it?
Thread poster: Przemyslaw Podmostko

Przemyslaw Podmostko  Identity Verified
United Kingdom
Local time: 21:39
English to Polish
+ ...
Nov 29, 2008

Hello!

I'm a beginner user of Trados (actually I like SDLX better). I just wanted to ask you colleagues what you do to increase usability of Translation Memories created by you, meaning what do you do in order for the TM to automatically translate as many phrases and words as possible?

E.g. is it possible to use SDLAlign to align two translations in a way that the TM is able to extract as many words and phrases as possible, and without doing it manually? (separating segments of a 20-pages word document will be a gruelling process).

Or do you just add words to the TermBase?

And if your language has declination (cases) do you add a phrase/word in all cases to the TermBase to improve finding matches or do you just decrease the percentage of fuzzy match probability?

Thank you in advance for any suggestions!

Kindest regards
Przemyslaw Podmostko


Direct link Reply with quote
 

Stanislav Pokorny  Identity Verified
Czech Republic
Local time: 22:39
English to Czech
+ ...
Comments Nov 29, 2008

Hi Przemyslaw,
I don't want to sound rude but this is a technical forum used by most ProZians to troubleshoot technical issues. Moreover, you can hardly expect any freelancer/competitor to reveal his/her know-how in organizing files etc.

Let me mention some comments though:

E.g. is it possible to use SDLAlign to align two translations in a way that the TM is able to extract as many words and phrases as possible, and without doing it manually? (separating segments of a 20-pages word document will be a gruelling process).


WinAlign is used to create TMs from previous translations, not to extract terminology (that's what TermExtract is for). That's an important point to remember. The alignment process automation degree depends heavily on technical preparation of the documents aligned. If they are poorly formatted, you can expect 100 % of work to be done manually (really a painful thing).


Or do you just add words to the TermBase?
And if your language has declination (cases) do you add a phrase/word in all cases to the TermBase to improve finding matches or do you just decrease the percentage of fuzzy match probability?


If you are talking about MultiTerm, adding all word forms (such as cases in Polish or Czech - in my case) won't help increase fuzzy matching. If you're using active terminology recognition, MultiTerm is only searching for index entries (i. e. for terms), not for their descriptive fields (i. e. word forms).

All in all, TMs are huge help but their capabilities are limited.


Direct link Reply with quote
 
hquynh trang
Local time: 03:39
English to Vietnamese
please help!!! Nov 29, 2008

i am a traslation student and i have just known Trados 3weeks ago. I really want to know how to use it.Unfortunately, i have searched for information about trados in internet but result is not so bright.I have the same question with Podmostko and be very grateful if anyone can send for me a tutorial file of trados.
my email: hquynhtrang@gmail.com
thanks!!!


Direct link Reply with quote
 

Heinrich Pesch  Identity Verified
Finland
Local time: 23:39
Member (2003)
Finnish to German
+ ...
These are two different things Nov 29, 2008

The termbase (Multiterm in Trados) is used to find known words or short phrases within a segment. That helps you to define terminology according to customer specification. So you do not expect the termbase to find normal everyday words but those that are significant for the task at hand.
The TM collects whole segments, mostly sentences or phrases, that can be used as such. The next time the same sentence comes up, the tm should find the previous translation. But if the author has put more or less tags into it, corrected a typo or exchanges a word for another you get a fuzzy match and have to decide, if your translation needs to be changed.

You can use the tm without a termbase.

Regards
Heinrich


Direct link Reply with quote
 

Przemyslaw Podmostko  Identity Verified
United Kingdom
Local time: 21:39
English to Polish
+ ...
TOPIC STARTER
Is there a way of optimising the TM? Nov 29, 2008

Heinrich Pesch wrote:

The termbase (Multiterm in Trados) is used to find known words or short phrases within a segment. That helps you to define terminology according to customer specification. So you do not expect the termbase to find normal everyday words but those that are significant for the task at hand.
The TM collects whole segments, mostly sentences or phrases, that can be used as such. The next time the same sentence comes up, the tm should find the previous translation. But if the author has put more or less tags into it, corrected a typo or exchanges a word for another you get a fuzzy match and have to decide, if your translation needs to be changed.

You can use the tm without a termbase.

Regards
Heinrich


Heinrich, I am aware that a termbase is not a general dictionary. My point was whether it is possible to optimise the TM or a termbase in order to increase probability of finding matches. Maybe someone has any tips developed by him/herself?

Thanks in advance!


Direct link Reply with quote
 

Ralf Lemster  Identity Verified
Germany
Local time: 22:39
English to German
+ ...
Check your installation Nov 29, 2008

HI Hquynh Trang,

i am a traslation student and i have just known Trados 3weeks ago. I really want to know how to use it.Unfortunately, i have searched for information about trados in internet but result is not so bright.I have the same question with Podmostko and be very grateful if anyone can send for me a tutorial file of trados.

An interactive tutorial is installed with the software - may I suggest to check that, as well as the documentation provided alongside.

Best regards,
Ralf


Direct link Reply with quote
 

ViktoriaG  Identity Verified
Canada
Local time: 16:39
English to French
+ ...
SDL Trados was designed for everything but finding matches Nov 29, 2008

I know, that's a bold statement. However, the sad truth is that if you really want to leverage your TMs and TDBs, you will need to use tools that don't come with the SDL seal of approval.

There are two (free) tools I use to furnish my bilingual corpora for translation. There is a third one that is quite expensive, but you may not need it. The two free tools I use are AntConc and XBench. The former is, to me, the most efficient term extraction tool (monolingual only) and the latter lets me search an unlimited number of bilingual files (TMs, TDBs, segmented Word files, TTX files, ITD files, etc.) simultaneously and, as long as the string you search for is there, it will find it (it has extensive search features). The third - quite expensive - tool is Nuance OmniPage, for OCR.

I create two types of resources: TMs and TDBs. For the creation of the former, I align various texts of which I know they contain the types of strings that will come in handy in my work and that they are reliable. A few examples are texts from government websites and texts from international organizations. You have to be very careful with the texts you pick - it is not because it is an official translation that it is necessarily a high-quality one. Since I am good at using OCR technology, I always try to get the PDF versions of these files for OCR processing. Why? Because when you can lay your hands on PDFs, the chances are much higher that the formatting between the two versions is exactly the same, which means that the OCR output will be much more suitable for alignment than the online copied-pasted text, the Word document, or pretty much anything else. Then, I just align these documents and store the TMs thus created for later use.

To create TDBs, I take the source text, save it as a text file and open it in AntConc. Without getting into much detail, let me just say that AntConc allows me to create a list of frequent words and word strings. I then dump that list into a text file, one term per line. Then, I translate the terms and add them to the text file, which transforms the text file into a tabbed text file. Then, I can import that into MultiTerm and I now have a brand new TDB. I don't believe it's any significant use to add whole sentences or phrases to TDBs. I don't have a use for that anyway, since XBench will find such phrases in the TM files.

The above answers the first question, which was how to gather text resources and create bilingual resources based on them. The next question is how to leverage all that.

Once again, I will not get into much detail about it, so I'll just say that XBench will let you search bilingual files quickly and efficiently. It will find things that SDL software will not. It also has very useful QA features, but that isn't the topic here. However, there is a key to using XBench: the organization of your bilingual resources. Actually, even if you don't use XBench, the organization of your resources is perhaps even more important than their content. I am against the method most people seem to favor, which consists in dumping all your smaller TMs into a single big momma TM. On the contrary, I like to keep smallish TMs whose subject matter, origins and, where applicable, client, are well identified, so I can mix and match them for each project. In XBench, you can specify an order of priority for your resources. XBench also displays information on the resource next to each result, so you always know which resource the search result comes from. If you use text fields or attribute fields, XBench also displays them. Which brings me to the importance of such fields.

Even if you don't use any external tools and stick with SDL products only, using fields allows you to distinguish between results in a concordance search. A good example is when you work on a project that was divided into chapters. Each chapter has its own document. If you use fields to identify TUs in the TM to each chapter, then, when you do a concordance search, you will be able to tell right away which chapter that result comes from. Similarly, you can use fields to label a TM (or part of it) with a subject, which will help you later when you mix several TMs together for a project. In my case, I have a dozen different subject fields, and when I want to work on an environmental translation, I just need to export all TUs labeled "Environment" to create a project TM. Of course, you can use fields to name clients, so, when results pop up in the concordance window, you see right away which client each result belongs to, and can take decisions based on such and such client's term preferences.

So, my advice to you is this:

1. Never stop gathering useful bilingual corpora and transforming them into translation resources (TMs, TDBs).
2. Keep 'em separated - this will allow you to custom mix and match resources for each translation project.
3. Learn to use text fields and attribute fields, and use them even when you don't think they will be useful.
4. Learn to use XBench and use it whenever you can. It will save you a lot of time and frustration.

All the best!


Direct link Reply with quote
 

Przemyslaw Podmostko  Identity Verified
United Kingdom
Local time: 21:39
English to Polish
+ ...
TOPIC STARTER
Thank you Viktoria! Nov 29, 2008

Thank you very much for such an exhausive post! I will sure check out those tools as soon as I can!

Direct link Reply with quote
 

Spiros Doikas  Identity Verified
Local time: 23:39
Member (2002)
English to Greek
+ ...
Keep reading and searching Nov 30, 2008

this forum. You will be amazed with the number of tips you can find.

Direct link Reply with quote
 

ViktoriaG  Identity Verified
Canada
Local time: 16:39
English to French
+ ...
One good resource Nov 30, 2008

hquynh trang wrote:

...very grateful if anyone can send for me a tutorial file of trados.


Try the Courses section (link in the left column) of this site: http://ecolotrain.uni-saarland.de/index.php?id=702&L=1


Direct link Reply with quote
 


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


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

Translation Memory - how to make the best of it?

Advanced search


Translation news related to SDL Trados





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 »
Déjà Vu X3
Try it, Love it

Find out why Déjà Vu is today the most flexible, customizable and user-friendly tool on the market. See the brand new features in action: *Completely redesigned user interface *Live Preview *Inline spell checking *Inline

More info »



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