Pre-translate doesn't pre-translate
Thread poster: Samuel Murray
Samuel Murray
Samuel Murray  Identity Verified
Netherlands
Local time: 02:52
Member (2006)
English to Afrikaans
+ ...
Aug 8, 2022

Hello everyone

This segment clearly has a 71% fuzzy match:

Capture-2

...but if I pre-translate using these settings:

Capture-1

...nothing happens. It doesn't insert the match for me.

Can you think of any setting that I may have accidentally set that would result in this behaviour? It's the second project today where this thing happens, from two different clients. The feature used to work just fine, for both clients. These are checked-out projects, by he way.

Thanks
Samuel

[Edited at 2022-08-08 16:14 GMT]


 
Thomas T. Frost
Thomas T. Frost  Identity Verified
Portugal
Local time: 01:52
Danish to English
+ ...
Can't think of anything Aug 8, 2022

I would report it as a bug to memoQ support (you don't need to have a current support and upgrade agreement to report bugs).

 
MollyRose
MollyRose  Identity Verified
United States
Local time: 19:52
English to Spanish
+ ...
happens a lot to me Aug 8, 2022

I just ignore it. When I confirm a TU, it populates the next one if Pretranslate didn't already.

I have found that if I have more than one TU with the same source, it doesn't populate, I guess because memoQ doesn't know which one to choose.


 
Samuel Murray
Samuel Murray  Identity Verified
Netherlands
Local time: 02:52
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
Not auto-propagation Aug 9, 2022

MollyRose wrote:
I just ignore it. When I confirm a TU, it populates the next one if Pretranslate didn't already.

I think you're confusing pre-translation with auto-propagation. Pre-translation is something that is done before you start translating and it's done automatically on all segments all in one go. It inserts the best fuzzy match from the TM above the given threshhold into all segments that have such fuzzy matches, all in one go. It also updates the fuzzy match percentages for all such segments (which is what I'm looking for, since I can then filter the segments on whether they have a fuzzy match from the TM).

My best suspicion is that there is some kind of way for the client to limit the pre-translation to a certain threshold (e.g. 99%)... possibly to avoid translators accidentally neglecting to edit low fuzzy matches.


Adieu
 
MollyRose
MollyRose  Identity Verified
United States
Local time: 19:52
English to Spanish
+ ...
difference Aug 9, 2022

Yes, I know, Samuel. I was responding to what you posted: the same thing happens to me a lot (when I pretranslate). Some segments fill and others don't. I have learned to not let it annoy me, because when I confirm a TU, it populates the next one if Pretranslate didn't already.

I can see where it could be a problem, though, if you need statistics at the beginning (between pretranslate and then adjusting TUs and doing new translation).


 
Johnny Leoni
Johnny Leoni
Ukraine
Local time: 03:52
English to Ukrainian
Several matches Aug 12, 2022

It really could be a problem with several matches with the same percentage, like on your screen.
If it is possible, try to exclude other TMs or strings and leave only one, the most suitable, and try to pre-translate the file again.


 
Hung Mai
Hung Mai
Vietnam
Local time: 07:52
English to Vietnamese
Are you working with a server-based TM? Aug 12, 2022

Samuel Murray wrote:

Can you think of any setting that I may have accidentally set that would result in this behaviour?


I encounter this issue quite often when working with a server-based TM in checked-out projects, I guess the pretranslate option is disabled on the TM server side (just like in some projects, I can't even lock strings). My suggestion is that you should contact your clients to ask for support on this option or just go along with the auto-propagation or create a local TM to store your translation for pretranslate option.


 
Samuel Murray
Samuel Murray  Identity Verified
Netherlands
Local time: 02:52
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
@Hung Aug 12, 2022

Yes, this is a checked-out project, so the TMs are online.

 
LEXpert
LEXpert  Identity Verified
United States
Local time: 19:52
Member (2008)
Croatian to English
+ ...
"Only Unambiguous Matches" Aug 12, 2022

MollyRose wrote:

I just ignore it. When I confirm a TU, it populates the next one if Pretranslate didn't already.

I have found that if I have more than one TU with the same source, it doesn't populate, I guess because memoQ doesn't know which one to choose.


Do you have "Only Unambiguous Matches" checked on the Pretranslation menu? I recently had a client with this issue, and I suspect was because they had this option checked. I had no problem pre-translating in the same project despite multiple 101/100% matches for the same segment (because I had revised my own translation several times). MemoQ sensibly simply took the newest match.


 
Samuel Murray
Samuel Murray  Identity Verified
Netherlands
Local time: 02:52
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
@Lexpert Aug 12, 2022

LEXpert wrote:
Do you have "Only Unambiguous Matches" checked on the Pretranslation menu?

Thanks. That option only becomes available if I select 100% matches. In my case, I want "Any match".

Also I don't care about whether the match is old or new -- it is the client's own responsibility to ensure that their TM contains only translations that they want me to re-use. If their TM contains two matches with identical match percentages, then I'll just use whatever one MemoQ chooses to insert.


Grigori Gazarian
 
Adieu
Adieu  Identity Verified
Ukrainian to English
+ ...
One problem Aug 12, 2022

71% is NOT a match. That's a long clause of a sentence that matches with something entirely different after the comma, or a part of a fully matching paragraph that was segmented differently in another file.

In MemoQ's opinion, it is inferior to MT and only a part of the segment matches.

It SHOULDN'T populate.

And yes, afaik the PM of a server project has a rather wide range of controls to lock you out of certain TM-related features, and anti-abuse measures
... See more
71% is NOT a match. That's a long clause of a sentence that matches with something entirely different after the comma, or a part of a fully matching paragraph that was segmented differently in another file.

In MemoQ's opinion, it is inferior to MT and only a part of the segment matches.

It SHOULDN'T populate.

And yes, afaik the PM of a server project has a rather wide range of controls to lock you out of certain TM-related features, and anti-abuse measures like keeping you from changing TM settings are usually applied.


[Edited at 2022-08-12 22:53 GMT]
Collapse


 
Samuel Murray
Samuel Murray  Identity Verified
Netherlands
Local time: 02:52
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
No, it's definitely a match Aug 13, 2022

Adieu wrote:
71% is NOT a match. That's a long clause of a sentence that matches with something entirely different after the comma, or a part of a fully matching paragraph that was segmented differently in another file.

How useful a 71% match is to a translator depends on a variety of factors, e.g. if the translator is only interested in speed increase or if the translator is trying to maintain consistency with the TM, or e.g. if the translator is able to easily see the source text differences, in order to quickly judge which part of the match is the useless part.

In the past, MemoQ has matched all the way down to 60% matches, which I prefer, since even a 60% match can be useful if the sentence is long enough. And if the setting in pre-translate is "Any TM match" then 71% should definitely qualify.

In this particular case, the current source text is:
If you have any questions, have a change in your health, or need to reschedule your visit, please call us:
and the match's source text is:
If you need to reschedule, have any questions, or experience a change in your health, please call us:

In MemoQ's opinion ... only a part of the segment matches.

Yes, that is what a fuzzy match is. Only part of the segment matches.

[Edited at 2022-08-13 19:46 GMT]


Stepan Konev
 
Stepan Konev
Stepan Konev  Identity Verified
Russian Federation
Local time: 03:52
English to Russian
Minimum match threshold = 40%. Any match = 50%+ Aug 13, 2022

Samuel Murray wrote:
"Any TM match" then 71% should definitely qualify.
Not only 71%. You can set the minimum match rate as low as 40% (Translation Settings - Automatic lookup and insertion) and they have to be populated indeed.
The memoQ manual reads:
Low fuzzy (50%-74%): In average-length or longer segments (8-10 words or more), the difference is more than two words. In pre-translation, "any match" means all sorts of fuzzy matches together, and they start at 50% (by default, because that can also be configured).


 
Stepan Konev
Stepan Konev  Identity Verified
Russian Federation
Local time: 03:52
English to Russian
60% for 9.12 Aug 13, 2022

Ok, the other version manual reads Any TM or corpus match: Always pre-translate a segment if there is a match, no matter the match rate. In fact, it must still be above minimum match threshold (60% by default).
Probably you should check your pre-translate settings. Open the Resource Console. Choose TM settings. Select your TM from the list, then click Edit. Clone it if requested. Check out your "Minimum match threshold". You mentioned that you use your client's online TM. Pro
... See more
Ok, the other version manual reads Any TM or corpus match: Always pre-translate a segment if there is a match, no matter the match rate. In fact, it must still be above minimum match threshold (60% by default).
Probably you should check your pre-translate settings. Open the Resource Console. Choose TM settings. Select your TM from the list, then click Edit. Clone it if requested. Check out your "Minimum match threshold". You mentioned that you use your client's online TM. Probably they have some different/non-default settings.


[Edited at 2022-08-13 20:52 GMT]
Collapse


 


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


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

Pre-translate doesn't pre-translate






Wordfast Pro
Translation Memory Software for Any Platform

Exclusive discount for ProZ.com users! Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value

Buy now! »
Anycount & Translation Office 3000
Translation Office 3000

Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.

More info »