| Pages in topic: [1 2] > |
I'm new - question about TM Thread poster: Hilde Vermee (X)
|
|---|
Hilde Vermee (X)
Belgium
Local time: 22:49
English to Dutch
+ ...
Hi all,
I'm new to Trados and I'm trying to organize my workflow. This might be a stupid question, but do I create a new TM for every file to be translated, or do I create a TM for every client (= translation agency) ?
Thanks for your help
| | | |
Daniel Šebesta 
Czech Republic
Local time: 22:49
Member (2007)
English to Czech
+ ...
If you specialize in only one field and one language pair, you might want to create only one TM for all your jobs. Agencies will normally provide you with a TM for each of their clients. Creating a TM for an agency doesn't make much sense if they send you projects for different end clients.
You certainly don't create a TM for each file. Why would you do that? You wouldn't be able to profit from the repetitions and matches.
HTH,
Daniel
| | | |
Hilde Vermee (X)
Belgium
Local time: 22:49
English to Dutch
+ ...
TOPIC STARTER
This what I wanted to know, your answer is very helpful Daniel, thanks!
| | | |
| My recommendations for Trados TM management | Sep 6, 2008 |
I think there is a lot of value in keeping all or most of your TM material in one place to get the most out of the concordance function. Doing so over many years has been a great benefit to me. In some cases you might want to work with a supplied TM and use yours as the reference TM or simply clean your work into your master TM later.
However, with Trados Workbench, I think it is important to use the attributes for distinguishing customers and subject areas. Doing so is useful in ma... See more I think there is a lot of value in keeping all or most of your TM material in one place to get the most out of the concordance function. Doing so over many years has been a great benefit to me. In some cases you might want to work with a supplied TM and use yours as the reference TM or simply clean your work into your master TM later.
However, with Trados Workbench, I think it is important to use the attributes for distinguishing customers and subject areas. Doing so is useful in many ways, including:
1) the ability to create filtered exports for your work or to give to your customer
2) understanding the context of matches you find using the concordance function
#1 has come in useful a number of times with clients who are not terribly well organized and tend to lose large amounts of TM material or forget to clean jobs into the end customer database. For these people I create an export a few times a year (when I notice missing content) and send it on to keep things consistent if they have to use another translator for update work. ▲ Collapse
| | |
|
|
|
Parrot
Spain
Local time: 22:49
Spanish to English
+ ...
| TM per end client | Sep 6, 2008 |
A TM per end client has the advantage of guaranteeing non-disclosure. Remember, you could get the same end-client from different agencies.
Some TMs may also be created per specialist area (civil law contracts based on standard models, such as insurance contracts, for example).
| | | |
| Non-disclosure | Sep 6, 2008 |
Parrot wrote:
A TM per end client has the advantage of guaranteeing non-disclosure. Remember, you could get the same end-client from different agencies.
If you use the attributes properly, that's not an issue. For example, I indicate agency, end customer and subject area. If/when I see a match I can decide what to do about it. Mostly the content shows up in concordance searches for specific terms anyway, where non-disclosure is barely relevant if at all.
There is nothing wrong with taking full advantage of one's own work as a reference. There may be many cases where you must use your professional judgment about how to apply the information, but looking at something in a TM is morally/ethically no different than simply remembering it. If my brain worked in a way that allowed me to recite thousands of pages after a brief glance at them, would it be in any way unethical to use this ability to my professional advantage? Of course not. Would I have to exercise proper discretion with what I remember? Of course.
If you do filtered exports for other people as I described, however, you had better be very, very sure you have set up your attribute definitions properly and used them every time. Otherwise intellectual property belonging to one customer might be passed on to another, and that would be very wrong (despite some of the half-baked contentions in the recent threads on a "global TM"). Parrot's suggestion protects you from this risk of course.
| | | |
Stuart Dowell
Poland
Local time: 22:49
Member (2007)
Polish to English
+ ...
Hi Kevin,
Does it mean that you maintain a single master TM which you use as a reference with each project?
Are there any issues with the size of the TM after having been building it for so many years?
Thanks
Stuart
| | | |
| Performance issues | Sep 6, 2008 |
stuart dowell wrote:
Does it mean that you maintain a single master TM which you use as a reference with each project?
Are there any issues with the size of the TM after having been building it for so many years?
Not that I've noticed. Every few years I upgrade my hardware; I don't know what the situation would be if I tried to use the laptop I had back in 2000.
I do in fact maintain client-specific TMs and subject-specific TMs for both Trados and DV for convenience in some cases (and scenarios where I really want to cut down on the number of concordance hits), but even in these cases I'm writing concurrently to the master TM/termbase or adding new content to it later.
There are a few huge Trados TMs that are supplied to me by clients, which are bigger than my compilations so far (or were last time I bothered to compare), and I don't see any particular performance problems when using those. My latest hardware is nothing special either - a cheap Toshiba laptop I picked up for under € 1000 two years ago. At least half the users here are probably working with something better.
This "big mama, big papa" strategy gets discussed among Déjà Vu users a lot. That particular software can even control the use of terms and segments in its assembly function based on client classifications (as well as based on subject area). With Trados the attribute information is more passive except when used as an export filter, but it's still very useful.
One obvious point I didn't mention before: I never add client-supplied TMs to my TMs. Once I have used/reviewed it in a translation and agreed with its content and format, some of it may make it into my reference TM, but I don't consider it ethical or wise to mass-load these resources into my master TM. Much of what I am supplied simply doesn't meet my quality standards, so even if it were legal or ethical, I wouldn't pollute my resources with it. Stuff like the EU TMs would get used strictly as a separate reference and then taken with a grain of salt in many cases. I do not use a TM for Client A for lookups when I am doing work for Client B except in rare cases such as where A or B supplies the other and such use is approved. (It does happen.)
[Edited at 2008-09-06 12:20]
| | |
|
|
|
| Avoid database corruption in SDL Trados TMs | Sep 6, 2008 |
First of all: Thanks to Hilde for asking a very relevant question. I assume the answers exceed her expectations and ability to implement them as a newby, but we're all here to learn something, so I'm glad the discussion got a momentum of its own.
Thanks to Kevin for this comment Kevin Lossner wrote:
One point I didn't mention before: I never add client-supplied TMs to my TMs.
Something I have not always been so consistently careful with, I must admit, but seeing as I have lost a big TM to database corruption after importing and that similar problems keep popping up now and again, always in connection with TM imports, I will take Kevin's advice to heart.
| | | |
| Backup, backup, backup | Sep 6, 2008 |
Marinus Vesseur wrote:
... I have lost a big TM to database corruption after importing and that similar problems keep popping up now and again, always in connection with TM imports...
I gripe a lot about Trados, but one rather wonderful thing about the product is the speed and ease with which one can back up the data from the TM to a text file. That process is brutally slow with my favorite product (Déjà Vu X).
I encourage you to back up both the actual files for the TM database (because I'm not sure if all the settings like segmentation rules are included in the text files - I've never bothered to check it out) and make text exports from the Workbench database as backups. If there is something corrupt in your data, you can always salvage most of your text export.
It also pays to learn to use a QA tool like ApSIC Xbench to clean up your TMs from time to time (as well as to do QA on uncleaned files). I am very much a beginner at this, but after spending half a day yesterday cleaning up 17,000+ translation units for a customer, I was amazed at how good it was for identifying problems with the content.
| | | |
| TMs per project | Sep 6, 2008 |
I create a TM for each project. My definition of project is a set of files that come from the same client - whether they are part of the same project on the client's end or not makes no difference. For example, one of my clients has me working on documents from the same author on the same subject, but they are different projects scattered over the year. I have one TM for all of these, which is practical because all the TUs are about the same subject and use the same terminology.
I r... See more I create a TM for each project. My definition of project is a set of files that come from the same client - whether they are part of the same project on the client's end or not makes no difference. For example, one of my clients has me working on documents from the same author on the same subject, but they are different projects scattered over the year. I have one TM for all of these, which is practical because all the TUs are about the same subject and use the same terminology.
I recommend you learn to use text fields in your CAT tool. What I do is create a TM and set the following fields: Client, Subject, Project and Document. Each time I reuse the TM, I set the variables for each document I translate. So, if I get a set of three files for translation, the labels attached to each TU look like this:
First document: Client A, Subject A and B, Project A, Document A
Second document: Client A, Subject B, Project A, Document B
Third document: Client A, Subject A, Project A, Document C
For environmental reports, I use several Subject fields, such as Environment, Report and Soil contaminants. Some of the documents within the same project will deal with only one of these, while others may deal with any combination or all of these. For the project field, I usually use the PO number given to me by the client for the project, or a short description of the project (e.g., Soil contaminants Quebec City). As for the document field, I simply copy the document title.
This allows me to keep large TMs that are very practical, but then I can also later export part of that large TM based on criteria like the subject, for example.
If you use text fields, you can build a few large TMs instead of many small ones which will not offer you the same benefit, without worrying that the TM gets too big for you to identify which TU concerns which client, subject, etc.
I suggest to learn to use text fields and always use them before you get too used to creating TMs in another way. Since I have figured this out, I always know exactly how my TMs are structured, and I can customize them to my heart's content.
One last thing: figure out a TM naming convention. You will see that as you create new TMs, you will get to a point where you will have dozens, or even hundreds, and if you don't have a sound naming convention for them, you can easily mistake one for the other. ▲ Collapse
| | | |
| Based on my experience, I'd advise keeping a TM per end-client. | Sep 6, 2008 |
Clients can be bizarrely or legitimately particular about terminology & usage. I had one major client who insisted that (French) "conforme" be translated as "complies".
Actually, "complies" is the term used in legal and regulatory matters but for technical stuff the ISO uses "conforms" (You do a "compliance report" for a financial transaction, a "conformity report" for a new lawn mower.)
The client's insistance can also be legitimate: One client had a history of legal docs that refe... See more Clients can be bizarrely or legitimately particular about terminology & usage. I had one major client who insisted that (French) "conforme" be translated as "complies".
Actually, "complies" is the term used in legal and regulatory matters but for technical stuff the ISO uses "conforms" (You do a "compliance report" for a financial transaction, a "conformity report" for a new lawn mower.)
The client's insistance can also be legitimate: One client had a history of legal docs that referred to "vendeur" as "seller" rather than "vendor" and wanted to keep it consistent.
Similarly, US-Can English is different from UK English, and increasingly different from the EU English invented by EU regulatory bodies. "Leverage/gearing", "power plant/station", "highway/motorway/autoroute", "enterprise/undertaking", "mutual fund / investment fund / collective investment", etc, etc....
(Interestingly, no CAT system offers "EU regulatory" as a type of English although it clearly now exists.)
My lovely wife's an IT consultant and has built me a database to house all my TMs and added a field so I can identify any particular term or segment as a "must" for a particular client.
(Which I'm happy to share with anyone interested.)
Failing that, I suggest keeping your TMs specific to the end-client.
Good luck ▲ Collapse
| | |
|
|
|
Hilde Vermee (X)
Belgium
Local time: 22:49
English to Dutch
+ ...
TOPIC STARTER | Thank you all | Sep 8, 2008 |
Marinus Vesseur wrote:
First of all: Thanks to Hilde for asking a very relevant question. I assume the answers exceed her expectations ...
Yes indeed, and I thank you all very much !! I have read your answers and tips very carefully and they will help me to structure my TM's from the beginning. I will follow the advice of Viktoria and find out about text fields, and combine this with the advice of Kevin to have one big TM.
Starting to work with Trados is quite exciting and I'm glad that I can ask the advice of experienced users.
| | | |
| Same thing, really | Sep 8, 2008 |
Hilde Vermeeren wrote:
I will follow the advice of Viktoria and find out about text fields, and combine this with the advice of Kevin to have one big TM.
When I say "attributes" I mean the same thing as Viktoria when she says "text fields". It's hard to avoid falling into jargon sometimes - sorry.
Make sure you learn how to designate a second TM as your reference TM:
Options menu -> Translation Memory Options... -> Concordance tab -> Reference (read-only) translation memory field
This will enable you to work with a client-supplied TM as your "main" TM while keeping your "big mama TM" in the background for term searching.
Another recommendation - which will probably incur boos and hisses - is to learn how to use MultiTerm (a lot of people HATE it). Use it both as an aid to translation and a tool for quality control of mandatory terminology. That's too big a subject to go into here (it is discussed in other threads). It's not as hard as many people think it is, and the long-term value of using this tool is very high. It will help to distinguish you from many other Trados users and to work faster and more accurately.
| | | |
| A TM per end client, AND... | Sep 8, 2008 |
Rimas Balsys wrote:
Based on my experience, I'd advise keeping a TM per end-client.
Absolutely. But sometimes you work for several agencies working for several divisions or branches of a big corporation. In that case, separating the TMs per end client, AND per agency, is advisable. Just for the purposes of privacy, keeping track of terminology differences between branches of one same corporation, etc.
| | | |
| Pages in topic: [1 2] > |
Login to reply/comment 
Translation news related to SDL Trados
