MemoQ preferring a 97% match over a 100% match
Thread poster: Edwal Rospigliosi

Edwal Rospigliosi  Identity Verified
Spain
Local time: 18:14
Member (2004)
English to Spanish
+ ...
Aug 15, 2010

Hi

I have this batch of documents with a lot of repetitions and almost-repetitions, such like these:

"1) Text to be translated"
"a) Text to be translated"
"Text to be translated"

And somehow, even after translating of them to 100%, when I work in a second document, I get this:

Translation results for "1) Text to be translated"

FIRST goes the translation of "Text to be translated", then "a) Text to be translated" and finally "1) Text to be translated". It goes even worse, in some cases the priority of the 100% match is 9 or 10, far below the individual glossary entries, therefore rendering a pre-translation useless.

Shouldn't MemoQ prioritize a 100% match over a 93% match?

Edit: I've noticed that MemoQ treats those 100% matches as "fragment assemblies" (in purple color) therefore sending them to the last place. That's even worse.

What is going on here? Any help is appreciated.

Regards,

[Editado a las 2010-08-15 10:16 GMT]


Direct link Reply with quote
 

Gergely Vandor
Hungary
Local time: 18:14
English to Hungarian
TM results are sorted my match rate, descending Aug 17, 2010

Hi Edwal,

The 97% hit should never be above a 100% hit, I've never heard of such a thing and I don't think it is likely. From your description, it seems that the 100% hit you are expecting simply doesn't exist in your TM.

Maybe you somehow didn't send that segment to the TM, perhaps you chose "Confirm without update" (Ctrl+Shift+Enter) instead of Confirm (Ctrl+Enter) by mistake.

Best regards,
Gergely Vandor

Kilgray


Direct link Reply with quote
 

Edwal Rospigliosi  Identity Verified
Spain
Local time: 18:14
Member (2004)
English to Spanish
+ ...
TOPIC STARTER
Incredible but true Aug 17, 2010

What I finally did was pretranslating to 100%, with "assemble" if no match. That way, the text body was merged with the numbering and everything "matched" at the end.

Direct link Reply with quote
 

Pablo Bouvier  Identity Verified
Local time: 18:14
German to Spanish
+ ...
MemoQ preferring a 97% match over a 100% match Aug 17, 2010

Edwal Rospigliosi wrote:

Hi

I have this batch of documents with a lot of repetitions and almost-repetitions, such like these:

"1) Text to be translated"
"a) Text to be translated"
"Text to be translated"

And somehow, even after translating of them to 100%, when I work in a second document, I get this:

Translation results for "1) Text to be translated"

FIRST goes the translation of "Text to be translated", then "a) Text to be translated" and finally "1) Text to be translated". It goes even worse, in some cases the priority of the 100% match is 9 or 10, far below the individual glossary entries, therefore rendering a pre-translation useless.

Shouldn't MemoQ prioritize a 100% match over a 93% match?

Edit: I've noticed that MemoQ treats those 100% matches as "fragment assemblies" (in purple color) therefore sending them to the last place. That's even worse.

What is going on here? Any help is appreciated.

Regards,

[Editado a las 2010-08-15 10:16 GMT]


Probably MemoQ do not consider numbers or individual letters and symbols at the beginning of the sentence as an element which belongs to the whole sentence, since normally initial numbers, initial letters or initial symbols are not translated and therefore they are not included in the comparison with other segments. I'm not sure, but I fear that this is a configuration issue because Kilgray usually does not give a stitch without a thread. Why you do not ask Kilgray's support directly?



[Edited at 2010-08-17 20:34 GMT]


Direct link Reply with quote
 

Gergely Vandor
Hungary
Local time: 18:14
English to Hungarian
support is here:) Aug 18, 2010

I don't think Edwal would get a better reply from our support than the one I gave above. To me it seems he is expecting a segment to be in his TM, while that segment is not there.

If he comes from a Trados background, there could be an explanation. For Trados (at least the old one, and with default settings, AFAIK), these two segments are "identical", and the second one will get a 100% match after the first one is translated, with the number automatically substituted:

"1) Text to be translated"
"2) Text to be translated"

This is good in this specific case, but can lead to a host of problems in general. memoQ, on the other hand, gives you a 100% match only if there is an actual exact match in your TM. Once again, Edwal's TM most probably didn't contain the exact same segment segment that he was translating, at least that is the only simple explanation I can give without having access to his TM and the document in question.

Gergely Vandor
Kilgray


Direct link Reply with quote
 

Grzegorz Gryc  Identity Verified
Local time: 18:14
French to Polish
+ ...
Numeral substitution Aug 18, 2010

Gergely Vandor wrote:

If he comes from a Trados background, there could be an explanation. For Trados (at least the old one, and with default settings, AFAIK), these two segments are "identical", and the second one will get a 100% match after the first one is translated, with the number automatically substituted:

"1) Text to be translated"
"2) Text to be translated"

This is good in this specific case, but can lead to a host of problems in general.

Seconded.
E,g, the Trados automatic numeral substutution is dangerours and absurd in Polish and other Slavic languages.
I.e. if we have sentences like:
I have 2 cats.
I have 5 cats.
the second one will receive a 100% match which is gramatically incorrect (the Slavic languages maintain vestiges of the ancient quaternary system and, e.g., the "logical" Nominative/Acccusative pl. is applied only to x2-x3-x4, the x5-x6-...-x1 sequences need Genetive pl.
I.e.
Mam 2 koty.
Mam 5 kotów.
It's not only the Trados problem, e.g. DVX acts in a similar way.

memoQ, on the other hand, gives you a 100% match only if there is an actual exact match in your TM.

I can but second this approach.

I didn't manage to reproduce the behaviour Edwal adresses.
BTW, MQ tries to perform some intelligent actions e.g. it tries to apply the initial upper case (which is wrong for the Edwal's example), so maybe in some special situations the algorithm is fooled.
You should send your real exaple to Gérgely.

Cheers
GG


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 »

MemoQ preferring a 97% match over a 100% match

Advanced search






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 »
WordFinder Unlimited
For clarity and excellence

WordFinder is the leading dictionary service that gives you the words you want anywhere, anytime. Access 260+ dictionaries from the world's leading dictionary publishers in virtually any device. Find the right word anywhere, anytime - online or offline.

More info »



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