TM Creation: Client vs Field Approach
Thread poster: Pavel Tsvetkov
| Client approach + termbase || Jan 28, 2009 |
From my experience of over a decade using Trados, what makes more sense and is safer is to have one memory per END CLIENT. I mean, if you work for an agency, it's safer to use one memory for each of the end clients of your agency customer, as you might get to work for a same end customer (a large corporation with several BUs or different locations in your country) via different agencies.
A memory per end client has the following advantages:
- You avoid the risk of unadvertedly including the name of a competitor in a translation for some customer, something that would be absolutely disastrous.
- You can be sure that you always use the terminology and style preferred by your customer along time.
The main advantage of a field memory approach is availability of concordance information from translations done for several customers. However, I think that it's a lot safer to create a termbase and use it in your translations for the different clients and end clients. If you design your termbase well, you can also reflect the terminology preferences of each end client.
| || || |
| | TSP
Local time: 19:18
German to Italian
| Client TM + complete TM || Jan 28, 2009 |
I totally agree with Tomás. We use 1 TM / client where we include correct client and project attributes + 1 big DB which contains all the client TMs (exported and re-imported into the DB) for concordance searches.
| | xxxMarc P
Local time: 19:18
German to English
A separate TM for every job. If you manage your TMs properly, you can then choose which of your legacy TMs to use for each new project, whether on the basis of subject, customer, date, etc. or a combination of attributes.
Some CAT tools achieve the same thing by making use of the attributes within a TM, although if I understand it correctly, this feature still has the disadvantage of unnecessary information being loaded and (possibly) indexed.
| | ViktoriaG
Local time: 13:18
English to French
| I'm with Marc || Jan 28, 2009 |
I also create a new TM each time I get a job, even if the job comes from the same end client, within the same field, through the same outsourcer.
If you want to be able to mix and match your TMs at times using the client criteria and at other times using the subject matter criteria, the best option you have is most likely the use of attribute fields.
I use Trados. I have gotten into the habit of using attribute fields for every single job, no matter how small. I generally use four attribute fields: Client, Project, Subject and Document (the latter so I can precisly tell which document a particular TU comes from, in case a job was made up of several files). If I want to use a reference TM for a job, I can select the appropriate TUs based on the criteria I deem fit for a particular job.
If you take all your small TMs and dump them all into a big momma TM, you will be able to use that as your reference TM (as you find results in it, you will see the content of the attribute fields and will be able to pick the appropriate TU). But you can also simply load the big momma TM and then export all TUs that match a set of attribute field based criteria you define. This way, you have complete freedom over the criteria you use.
In case you do have a big momma TM and it is becoming a bit too large for it to be efficient, then I would split it into a few smaller, subject matter based TMs.
If you use Trados, please read the relevant user guide sections on attribute fields. You may discover how handy these are - and they are useful for a bunch of other purposes as well, not the least QA.
| || || |