List of XTM XLF state-qualifiers and their meanings
Thread poster: Samuel Murray
Samuel Murray
Samuel Murray  Identity Verified
Netherlands
Local time: 00:55
Member (2006)
English to Afrikaans
+ ...
Aug 19, 2016

Hello

For a while now, I've been translating XTM XLF files via OmegaT, using the Okapi filter and non-sentence segmentation.

I have now discovered a quirk in XTM's XLF format, namely that populated fuzzy matches are marked as state="translated", even though it is an unconfirmed fuzzy match. Certain fuzzy matches are marked as state="translated" state-qualifier="fuzzy-match"... See more
Hello

For a while now, I've been translating XTM XLF files via OmegaT, using the Okapi filter and non-sentence segmentation.

I have now discovered a quirk in XTM's XLF format, namely that populated fuzzy matches are marked as state="translated", even though it is an unconfirmed fuzzy match. Certain fuzzy matches are marked as state="translated" state-qualifier="fuzzy-match", but none of the programs that I opened the XTM XLF file in (e.g. OmegaT, Trados 2015, Wordfast 4, Virtaal) recognise the "state-qualifier", and so all of those programs assume that all those unedited fuzzy matches are in fact correct translations.

I don't really need/use inserted fuzzy matches, because I generally have the TM from which the fuzzy match came from to begin with. So, for me it would be okay to simply blank out the target field of segments marked as populated fuzzies. I assume this means that I simply have to delete all target texts of segments that use state="translated" state-qualifier="fuzzy-match" from the XLF file, before loading it into the editor.

However, this discovery of state-qualifiers has led me to unearth more of them, e.g.:

state="translated" state-qualifier="fuzzy-match"
state="translated" state-qualifier="x-fuzzy-forward"
state="translated" state-qualifier="leveraged-inherited"
state="translated" state-qualifier="leveraged-tm"
state="translated" state-qualifier="exact-match"

...and I'm not sure what they mean. I mean, I don't know which of them relate to translations that I can trust, and which of them indicate that the existing translation is almost certainly incorrect.

Does anyone have a list of these state-qualifiers, and/or know what they mean?

Thanks
Samuel


[Edited at 2016-08-19 09:11 GMT]
Collapse


 
XTM Intl
XTM Intl
United Kingdom
Local time: 23:55
Work with TMs to see source differences, populate target with fuzzy matches only for work in XTM Aug 24, 2016

Dear Samuel,

In order to display source differences between the source text existing in the source file and source text of the applied fuzzy match you have to connect the translation memory from which these fuzzy matches come from when you work in Trados.

Working without source text differences displayed for fuzzy matches can lead to substantial quality issues or can greatly impact translation productivity. For this reason, XTM advises to work with the TM and without po
... See more
Dear Samuel,

In order to display source differences between the source text existing in the source file and source text of the applied fuzzy match you have to connect the translation memory from which these fuzzy matches come from when you work in Trados.

Working without source text differences displayed for fuzzy matches can lead to substantial quality issues or can greatly impact translation productivity. For this reason, XTM advises to work with the TM and without populating target text with fuzzy matches, especially as the feature to populate target text with fuzzy matches is disabled by default and will only facilitate the translation process in XTM Editor. On the other hand, XLF file which is not populated with fuzzy matches does not lose leverage because these TM matches are applied from the TM.

In such case, provided that you follow XTM advice, you do not need to analyse the structure of XTM XLF files to know which segments contain correct and which contain almost certainly incorrect translation.

However, below please see explanations of XTM state qualifiers:
state="translated" state-qualifier="fuzzy-match" — means that the highest match on the list is a fuzzy match
state="translated" state-qualifier="x-fuzzy-forward" — means that the highest match on the list is a fuzzy repetition
state="translated" state-qualifier="leveraged-inherited" — means that the highest match on the list is a 100% repetition
state="translated" state-qualifier="leveraged-tm" — means that the highest match on the list is a leveraged TM match (a 100% match from a TM)
state="translated" state-qualifier="exact-match" — means that the highest match on the list is an ICE exact match

We hope this will help you. Should you have any other questions, please contact XTM support directly at [email protected].

Best regards,
XTM Support
Collapse


 
Samuel Murray
Samuel Murray  Identity Verified
Netherlands
Local time: 00:55
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
Thanks Aug 24, 2016

XTM Intl wrote:
For this reason, XTM advises to work with the TM and without populating target text with fuzzy matches...


Unfortunately the fuzzy matches are inserted at the time that the job is assigned to me )-:

However, below please see explanations of XTM state qualifiers:


Thanks
Samuel


 
Stepan Konev
Stepan Konev  Identity Verified
Russian Federation
Local time: 02:55
English to Russian
fuzzy repetition Sep 1, 2019

Hi XTM Support,

Could you please elaborate on the diefference between 'fuzzy repetition', 'fuzzy macth', and '100% repetition'.
I understend the fuzzy macth concept, but no idea about fuzzy repetition. How can a repetition be other than 100%?
Thank you.


 
Samuel Murray
Samuel Murray  Identity Verified
Netherlands
Local time: 00:55
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
Internal fuzzy matches Sep 1, 2019

Stepan Konev wrote:
Could you please elaborate on the diefference between 'fuzzy repetition', 'fuzzy macth', and '100% repetition'.


From what I can see in the XTM user manual, and my experience with these terms in other CAT tools, a "fuzzy repetition" is nothing more than an internal fuzzy match. In other words, it's a sentence in the source files that is very similar to another sentence in the source files, regardless of whether there is a similar sentence in the TM. The XTM user manual appears to use the term "fuzzy match" for an external fuzzy match, i.e. a sentence in the TM that is very similar to a sentence in the source files. However, the XTM user manual does not use the terms consistently. In one place, for example, it refers to fuzzy repetitions as "document fuzzy matches". In one place it speaks of "exact matches" but elsewhere the same concept is referred to as "ICE matches", and what was normally called "100% matches" in most CAT tools are often called "leveraged 100% matches" in the XTM user manual. The term "100% repetition" does not occur in the XTM user manual, though I imagine it would simply refer to a repetition, i.e. a sentence in the source files that is identical to another sentence in the source files.


 
Stepan Konev
Stepan Konev  Identity Verified
Russian Federation
Local time: 02:55
English to Russian
Thank you Samuel Sep 1, 2019

Samuel Murray wrote:
From what I can see in the XTM user manual, and my experience with these terms in other CAT tools, a "fuzzy repetition" is nothing more than an internal fuzzy match.


You mean that conventional 'fuzzy match' relies on TM, while 'internal fuzzy match' relies on other similar sentences within a documnet even if they are not in TM yet, don't you? If so, it seems clear now. Thank you Samuel.
I have 2 pdf manuals, but none of them mention the 'fuzzy repetition' term.

(However I would still appreciate any sort of confirmation from XTM team, if they happen to come across this thread.)


 
Kjersti Farrier
Kjersti Farrier  Identity Verified
United Kingdom
Local time: 23:55
English to Norwegian (Bokmal)
+ ...
Where is the option to not populate target text with fuzzy matches May 9, 2020

[quote]XTM Intl wrote:

For this reason, XTM advises to work with the TM and without populating target text with fuzzy matches, especially as the feature to populate target text with fuzzy matches is disabled by default and will only facilitate the translation process in XTM Editor. On the other hand, XLF file which is not populated with fuzzy matches does not lose leverage because these TM matches are applied from the TM.

--------------

I am using Trados Studio 2015. I have an XTM project that I would like to work on offline. I have downloaded all the files and TM. My issue is that all the segments with fuzzy matches from the online files becomes translated in my file in Trados. These obviously needs to be checked as they are clearly incorrect.
Where can I find this option to not populate the target text with "incorrectly translated" fuzzy matches? It says that this feature is disabled by default, but that cannot be the case for me. I really hope that someone still reads this post, because it is getting old...

Many thanks,
Kjersti


 
XTM Intl
XTM Intl
United Kingdom
Local time: 23:55
Not populate target text with fuzzy matches - XTM May 9, 2020

Kjersti Farrier wrote:

I am using Trados Studio 2015. I have an XTM project that I would like to work on offline. I have downloaded all the files and TM. My issue is that all the segments with fuzzy matches from the online files becomes translated in my file in Trados. These obviously needs to be checked as they are clearly incorrect.
Where can I find this option to not populate the target text with "incorrectly translated" fuzzy matches? It says that this feature is disabled by default, but that cannot be the case for me. I really hope that someone still reads this post, because it is getting old...

Many thanks,
Kjersti


-------------------------------------

Dear Kjersti,

thank you for your post. I am Product Manager at XTM and would like to get back to you. And no worries, we monitor Proz and will get back to you any time you need help or want to share any feedback on our product.
Regarding your question, the setting controlling which TM matches should automatically populate the segments can be accessed only by the XTM Admin user of the XTM instance your project comes from. If you are only a linguist (translator, corrector roles, etc.) and have no access to the XTM configuration, you have to reach out to the Project Manager/anyone on your client side.

Just for you to know, the client can find this option under Configuration > Settings > Translation > TM tab > 'Fuzzy matches' section - Populate target with...
The option needs to be set to 'Never' in all fuzzy categories and this is the default setting in XTM.
This would apply to both online and offline translation.

I would like to check anyways if, in your case, the target segments are populated AND confirmed as well in Trados. The fact that you see fuzzy matches in the target segments in Trados could just be because the matches are actually coming from the .tmx file you have connected to your project. That's why you would see them, but in this case, they would not be confirmed (Trados segment status would be 'Draft' for fuzzy matches coming from the tmx file, and 'Translated' for the matches that were automatically or manually confirmed in XTM before generating the .xlf file).

So I would advise you to check with your client the System settings, but also to double-check if in Trados those matches are just coming from the .tmx and not the .xlf file itself.

As for the user manual, the most up to date version (12.3) can be downloaded either from the XTM UI (under the ? icon) or from our website https://xtm.cloud/documentation/
I can confirm that fuzzy repeats belong to the internal fuzzy match categories in XTM. While the standard fuzzy matches refer to the matches coming from the TM. Please refer to the XTM manual, page 294.

I hope this helps, otherwise feel free to reply to this post or reach out to Support at [email protected]

Sara Basile
Product Manager
XTM International Ltd.
www.xtm.cloud


 
Samuel Murray
Samuel Murray  Identity Verified
Netherlands
Local time: 00:55
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
I have a little script... May 9, 2020

Kjersti Farrier wrote:
Where can I find this option to not populate the target text with "incorrectly translated" fuzzy matches?


If I understand correctly, the option is available only to the client. The translator has no access to it.

I have a pair of scripts that takes the XTM XLIFF file and attempts to either blank out or mark up the fuzzy translations in the XLIFF file.

http://www.leuce.com/autoit/XTMunfuzzy.zip
(you need AutoIt installed)

Updated: I added a readme file to the scripts (same zip file), and the scripts work for me. But of course, I have only have limited XTM XLIFF files to test it on.



[Edited at 2020-05-09 21:34 GMT]


 
Kjersti Farrier
Kjersti Farrier  Identity Verified
United Kingdom
Local time: 23:55
English to Norwegian (Bokmal)
+ ...
I did wonder if I could edit the xlf file somehow May 9, 2020

Thank you, Samuel, for taking your time to respond.
I did wonder if I could somehow meddle with the .xlf file actually... but couldn't figure out how without a script. I need to handle another job first, but will take a look later.

Thanks again,
Kjersti


 
Samuel Murray
Samuel Murray  Identity Verified
Netherlands
Local time: 00:55
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
@Sara from XTM May 10, 2020

XTM Intl wrote:
state="translated" state-qualifier="leveraged-inherited" — means that the highest match on the list is a 100% repetition


I'm unclear about what "a 100% repetition" means. Does it mean that it is a repetition of another segment for which there is a 100% match? Or does it simply mean that it is a segment that is identical to another segment in the project?

In other words, if there are e.g. two identical segments that both have a fuzzy match in the TM, would they be labelled "leveraged-inherited"? Or, if there are e.g. two identical segments that both have no match in the TM, would they be labelled "leveraged-inherited"?


 
XTM Intl
XTM Intl
United Kingdom
Local time: 23:55
To Samuel: 100% repetition and state-qualifier="leveraged-inherited" May 12, 2020

Samuel Murray wrote:

XTM Intl wrote:
state="translated" state-qualifier="leveraged-inherited" — means that the highest match on the list is a 100% repetition


I'm unclear about what "a 100% repetition" means. Does it mean that it is a repetition of another segment for which there is a 100% match? Or does it simply mean that it is a segment that is identical to another segment in the project?

In other words, if there are e.g. two identical segments that both have a fuzzy match in the TM, would they be labelled "leveraged-inherited"? Or, if there are e.g. two identical segments that both have no match in the TM, would they be labelled "leveraged-inherited"?



--------------

Hi Samuel,
thank you for your question and apologies for my late reply.
'leveraged-inherited' (or 100% repetition, as we called it) just refers to an internal repetition, hence two (or more) identical segments in the project files.

Note that only the repeated segment(s) would have the state qualifier 'leveraged-inherited' and not the first occurrence of that segment.

In particular, to go back to your example, this is what happens:

Case 1: if there are e.g. two identical segments that both have a fuzzy match in the TM
- First segment has state-qualifier="fuzzy-match" [followed by match-quality="xx%" and the information about the original project where the fuzzy is coming from, with the original source and target match of that fuzzy]
- Second/repeated segment (and any following repetition) has

Case 2: if there are e.g. two identical segments that both have no match in the TM
- First segment has no state qualifier, because it is unmatched
- Second/repeated segment (and any following repetition) has

Hope it helps!
Best,

Sara Basile
Product Manager
XTM International Ltd.
www.xtm.cloud


 


There is no moderator assigned specifically to this forum.
To report site rules violations or get help, please contact site staff »


List of XTM XLF state-qualifiers and their meanings






Protemos translation business management system
Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!

The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.

More info »
Trados Studio 2022 Freelance
The leading translation software used by over 270,000 translators.

Designed with your feedback in mind, Trados Studio 2022 delivers an unrivalled, powerful desktop and cloud solution, empowering you to work in the most efficient and cost-effective way.

More info »