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

Giles Watson  Identity Verified
Italy
Local time: 23:48
Italian to English
In memoriam
Relevance is in the eye of the translator Sep 28, 2014

Meta Arkadia wrote:

Dominique Pivard wrote:
+1: I've never understood all the fuss about auto-assembly/fragment assembly


Just look upon it as a way to make the CAT tool provide you with the best suggestions from your resources, based on the tool's algorithms and your settings. A regular search will a show you all hits, AA will show you only the most relevant hit.



Relevant to what, though?

Unless you are translating very technical stuff, there will probably be a range of relevant options for any particular translation crux. All sorts of considerations that cannot be predicted by algorithms or settings - the source author's use of figurative language, prosody and so on - come into the decision. I want to see enough relevant instances from my TMs to inspire a new solution and in these circs, subsegment assembly is overkill.

I'm not knocking AA, which suits a lot of text types, including yours. It just doesn't work very well for the discursive translations I tend to get.


 

Meta Arkadia
Local time: 05:48
English to Indonesian
+ ...
It's an extra, Giles Sep 28, 2014

Giles Watson wrote:
I want to see enough relevant instances from my TMs to inspire a new solution and in these circs, subsegment assembly is overkill.

Whether you enable or disable AA, the search results are the same. So if you're not happy with AA suggestion, you can have a look at all the other search results, without extra effort. It's a bonus. A very useful bonus that can make your translation more consistent if needed, and can save you a lot of typing.

Cheers,

Hans


 

Giles Watson  Identity Verified
Italy
Local time: 23:48
Italian to English
In memoriam
I might have another look Sep 29, 2014

Meta Arkadia wrote:

Giles Watson wrote:
I want to see enough relevant instances from my TMs to inspire a new solution and in these circs, subsegment assembly is overkill.

Whether you enable or disable AA, the search results are the same. So if you're not happy with AA suggestion, you can have a look at all the other search results, without extra effort. It's a bonus. A very useful bonus that can make your translation more consistent if needed, and can save you a lot of typing.

Cheers,

Hans


I might have another look at it, then, Hans. To be honest, what saves me most typing at the moment is AutoSuggest.

Best,

Giles


 

Meta Arkadia
Local time: 05:48
English to Indonesian
+ ...
Auto-completion Sep 29, 2014

Giles Watson wrote:
To be honest, what saves me most typing at the moment is AutoSuggest.


But I'm sure you realise AutoSuggest also works according to something similar to Auto-Assembling. That is, that's the case for CafeTran's Auto-Completion. Auto-Completion suggests

• Words that you type in the target box
• Words in your User.dic, your personal list of words with unknown spelling
• Words in your Translation Memory
• Words in the target segments of your translation project
• Words at the target side of your glossary

It's "dynamic" in the sense there's a hierarchy (like words from the project file first), and that it suggests only words from the attached resources. Some CAT tools provide a static solution though, suggesting words from a pre-compiled list. You mentioned "AutoSuggest,' but I'm afraid I don't know to which CAT tool it belongs, and how it works.

Cheers,

Hans


 

Giles Watson  Identity Verified
Italy
Local time: 23:48
Italian to English
In memoriam
Trados AutoSuggest Sep 29, 2014

Meta Arkadia wrote:

t's "dynamic" in the sense there's a hierarchy (like words from the project file first), and that it suggests only words from the attached resources. Some CAT tools provide a static solution though, suggesting words from a pre-compiled list. You mentioned "AutoSuggest,' but I'm afraid I don't know to which CAT tool it belongs, and how it works.



AutoSuggest is a Trados (other CATs are available) plug-in that offers mouse-tip suggestions from the user's TMs. It's static in the sense that it doesn't update automatically so you have to regenerate it to add stuff but it works very well if you have sufficiently close-focused TMs.

DV's old AutoAssemble was a PITA because its suggestions took more time to verify and correct than translating from scratch. IIRC you had the choice of automatically populating your target window with what usually turned out to be rubbish or selecting the useful bits from another part of the screen.

Mouse-tip options, which I was leery of to start with but have since found to be very helpful, let you see straight away if the CAT has anything useful to offer. They save you a lot of typing time, are simple to insert and don't force you to move your eyes to another part of the screen.


 

Meta Arkadia
Local time: 05:48
English to Indonesian
+ ...
Mouse-tip? Sep 29, 2014

Giles Watson wrote:
DV's old AutoAssemble was a PITA because its suggestions took more time to verify and correct than translating from scratch.


I was quite happy with DV3's AA. It probably depends on your language pair. The disadvantage was - if I remember correctly - that it always inserted the AA suggestion in your target pane, and yes, sometimes it was more trouble re-arranging the subsegments than simply entering the translation. CafeTran offers you a choice for AA, either inserted in the target pane, or shown as a pop-up.

Giles Watson wrote:
Mouse-tip options, which I was leery of to start with but have since found to be very helpful, let you see straight away if the CAT has anything useful to offer. They save you a lot of typing time, are simple to insert and don't force you to move your eyes to another part of the screen.

I fail to understand what a mouse has to do with it. In CafeTran, you typ a few letters, the suggestions pop up, and you hit Return to enter the first suggestions, or a number for the next ones.


[image stolen from HansL's wiki]

The screenshot above shows clearly why I dared to compare CafeTran's Auto-Completion with Auto-Assembly. Based on the project file, the term used in segment 1 is suggested first. The preferred suggestion, because you only have to hit Return.

Cheers,

Hans


 

Giles Watson  Identity Verified
Italy
Local time: 23:48
Italian to English
In memoriam
OK then, "track-point tip". Sep 29, 2014

Meta Arkadia wrote:

I fail to understand what a mouse has to do with it.



"Mouse" is simply a generic term for the "pointing device of your choice". In any case, I prefer track-points


 

Michael Beijer  Identity Verified
United Kingdom
Local time: 22:48
Member (2009)
Dutch to English
+ ...
caret/cursor vs mouse pointer Sep 29, 2014

I think the term we are looking for is ‘caret’ or ‘cursor’. At least, that's where the auto-complete stuff shows up in CafeTran. My mouse pointer can be somewhere else completely.

See: http://www.computerhope.com/jargon/c/cursor.htm

caret/cursor + mouse pointer

Michael


 

Giles Watson  Identity Verified
Italy
Local time: 23:48
Italian to English
In memoriam
Cursor it is, then Sep 29, 2014

Thanks, Michael.

"Cursor" is fine (I'm less happy with "caret", which of course denotes this symbol: ^).

Getting back to the topic though, CafeTran's auto-complete looks very much like the Trados autosuggest function that is making my life a little easier, thanks to AS dictionaries generated from focused TMs.

I really must get round to road testing CT sometime soon


 

Michael Beijer  Identity Verified
United Kingdom
Local time: 22:48
Member (2009)
Dutch to English
+ ...
Proz webinar tomorrow on MT + freelancers Sep 29, 2014

Hi Giles,

I'd definitely recommend having a look at it. It's cheap, light, and very powerful.

Incidentally, and perhaps a little off-topic, but I will be talking about CafeTran a bit in the webinar tomorrow called ‘Machine Translation panel from a freelancer perspective’. I am one of the panellists. I'll probably be discussing how MT is integrated in CT, but might also touch upon other things such as auto-complete, etc.

info: ... See more
Hi Giles,

I'd definitely recommend having a look at it. It's cheap, light, and very powerful.

Incidentally, and perhaps a little off-topic, but I will be talking about CafeTran a bit in the webinar tomorrow called ‘Machine Translation panel from a freelancer perspective’. I am one of the panellists. I'll probably be discussing how MT is integrated in CT, but might also touch upon other things such as auto-complete, etc.

info: http://www.proz.com/virtual-conferences/632/program/9778

Michael
Collapse


 

Meta Arkadia
Local time: 05:48
English to Indonesian
+ ...
Dynamic Sep 30, 2014

Giles Watson wrote:
I really must get round to road testing CT sometime soon


If you rely on Auto-Completion, I suggest you do try CafeTran. In the example below, you can see I only attached a ProjectTM. Even after one segment, CafeTran already suggests the right word. Hit Return (Enter), and the term appears in your target pane, and CafeTran will suggest a second word (and a third, and... ad infinitum), taken from the only resource it can use in this case (which is wrong, of course, but you should just continue typing to skip it).



A "static" AutoSuggest can never accomplish this.

And again, this underlines how important organising and selecting your TMs is. Picking the right TM will increase the number of "Auto-Completion hits" immensely.

Cheers,

Hans

[Edited at 2014-09-30 00:58 GMT]


 

Giles Watson  Identity Verified
Italy
Local time: 23:48
Italian to English
In memoriam
Static vs dynamic Sep 30, 2014

Thanks for going to the trouble of posting the animation, Hans!

Most of the time, I want to browse legacy work for - perhaps half-remembered - past translation solutions to adapt them to a new context. Currently, Trados does a good job with a combination of concordance searches (highlight + F3) and the drop-down AutoSuggest cursor menu (© M. Beijer) proposing options from my AS dictionaries and MultiTerm termbases. "Noise" from inappropriate suggestions - the big drawback with MT -
... See more
Thanks for going to the trouble of posting the animation, Hans!

Most of the time, I want to browse legacy work for - perhaps half-remembered - past translation solutions to adapt them to a new context. Currently, Trados does a good job with a combination of concordance searches (highlight + F3) and the drop-down AutoSuggest cursor menu (© M. Beijer) proposing options from my AS dictionaries and MultiTerm termbases. "Noise" from inappropriate suggestions - the big drawback with MT - is kept to a minimum.

A static set-up is fine for this but I'm now getting curious to see what difference CT's dynamic approach will make.
Collapse


 

Meta Arkadia
Local time: 05:48
English to Indonesian
+ ...
Another GIF Sep 30, 2014

Giles Watson wrote:
A static set-up is fine for this but I'm now getting curious to see what difference CT's dynamic approach will make.

That should be easy to test, Giles. It seems AutoSuggest is usually based on term extraction of the DGT. So to test CafeTran, I attached the latest DGT (450,000 segments) to my "Giles Project." I'm learning a lot, because I thought Auto-Completion wouldn't work for the very first segment. It did.



Just to show off, I also performed a search in this pretty big TM. I must admit I assigned a generous 8 GB of RAM to Java. 4 GB would probably already do.

So you can easily test Auto-Completion by using the "raw" DGT as a resource. It should render at least as many Auto-Completion results as the extracted AutoSuggest term list. Without extra work.

Cheers,

Hans


 

Dan Lucas  Identity Verified
United Kingdom
Local time: 22:48
Member (2014)
Japanese to English
TOPIC STARTER
Fascinating stuff Sep 30, 2014

Just want to thank everybody for their contributions, especially Hans and Michael on this page for showing what is possible with a carefully tuned setup. For those who do use CAT this has been an excellent discussion.

Dan


 

Giles Watson  Identity Verified
Italy
Local time: 23:48
Italian to English
In memoriam
Fame at last! Sep 30, 2014

Meta Arkadia wrote:

It seems AutoSuggest is usually based on term extraction of the DGT. So to test CafeTran, I attached the latest DGT (450,000 segments) to my "Giles Project." I'm learning a lot, because I thought Auto-Completion wouldn't work for the very first segment. It did.



Fame at last!

Kidding aside, thanks for the new info, Hans. I mainly use AutoSuggest with dictionaries generated from my own wine and journalism TMs, both of which contain well over 100,000 TUs. Recently, I created two other AS dictionaries from the IATE IT-EN and EL-EN glossaries, which turn out to be useful for quite a lot of stuff.

Must find time soon for CafeTran, or as I am now going to have to call it, the "Hans Project".


 
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







CafeTran Espresso
You've never met a CAT tool this clever!

Translate faster & easier, using a sophisticated CAT tool built by a translator / developer. Accept jobs from clients who use SDL Trados, MemoQ, Wordfast & major CAT tools. Download and start using CafeTran Espresso -- for free

More info »
SDL MultiTerm 2019
Guarantee a unified, consistent and high-quality translation with terminology software by the industry leaders.

SDL MultiTerm 2019 allows translators to create one central location to store and manage multilingual terminology, and with SDL MultiTerm Extract 2019 you can automatically create term lists from your existing documentation to save time.

More info »



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