Verification issue: "Repeated word found" error for numbers despite "ignore numbers" being set in QA Checker settings

Hi

I get these errors,

Trados Studio error message showing repeated word error for numbers separated by spaces and a multiplication symbol.

despite these settings (to ignore numbers).

Trados Studio project settings window with 'Ignore numbers' option checked under 'Verification' settings.

Why is this happening? I don't even know why it's considered to be "repeating" as two spaces and a multiplication symbol separate the numbers.
Is this worth posting to "Ideas"?



Generated Image Alt-Text
[edited by: Trados AI at 4:29 PM (GMT 0) on 28 Feb 2024]
emoji
Parents
  • Ignoring the fact it should not be checking this if you unchecked it, the reason it's considered repeating is because they are repeated.  You only have one '92' in the source, but two in the target.

    Can I just ask if you are using the language specific check and perhaps this is why it appears to be ignoring your "uncheckedness"?

    Paul Filkin | RWS Group

    ________________________
    Design your own training!

    You've done the courses and still need to go a little further, or still not clear? 
    Tell us what you need in our Community Solutions Hub

  • Paul,
    "the reason it's considered repeating is because they are repeated".

    Okay... so you mean in the segment "The dog ran across the street" would be flagged because "the" repeats? What should happen is "The dog ran across the the street" should be flagged because "the" is repeated in succession. There is a character between "92" so it isn't repeated. 
    Anyway, to test your theory, I edited the source and added the number then re-verified it. Still gets flagged. One or two "92"s in the source is irrelevant.

    Trados Studio error list showing multiple 'Repeated word' warnings for segments with the number 92.

    Turning on tokenization of numbers for the top TM and redoing the verification still get's flagged.
    Close-up of Trados Studio segment with 'Repeated word' warning for the phrase '92 92' despite non-repeating context.

    But, just as I thought, if I had word ( or even a letter) between the 92, then the error goes away. Trados Studio error list with 'Repeated word' warnings, including one for a segment with the multiplication symbol between 92s.

    So why isn't a symbol [×], the SI symbol for multiplication, being recognized? That symbol not included in the script/code perhaps whereas letters are? 
    And why isn't "ignore numbers" working in the first place? 

    As for "Can I just ask if you are using the language specific check" I believe you are referring to a function of studio 2019 but we use the latest version of 2017. 

    emoji


    Generated Image Alt-Text
    [edited by: Trados AI at 4:30 PM (GMT 0) on 28 Feb 2024]
  • As for "Can I just ask if you are using the language specific check" I believe you are referring to a function of studio 2019 but we use the latest version of 2017. 

    Noted... good to clarify this in the intial post.

    In general I can't give you good answers for your questions for the following reasons:

    • number/measurement recognition in Studio needs work
    • we can see what is happenning wrt recognition in the source fairly easily (although not from your screenshots), but it's not so easy to understand what's happenning in the target.  For example:
      Screenshot of Trados Studio showing QA Checker 3.0 errors: repeated word found, number missing in target segment, and extra measurement in target segment.
      All of these segments have weird results (excuse the JP as I had to guess) but I'll use this last one to explain what I mean:
      - source clearly recognises three numbers
      - target sees two numbers and a measurement (92, 92 and 32 mm)
      - repeated word... Studio seems to smoking something here!

    I would normally suggest switching off the out of the box QA for this and use the Number Verifier as it does provide more granular control but then you lose the ability to use parts f the QA checker that are still useful.

    I'll share this with support, as usual, but there isn't a good workaround for this.  This part of Studio needs to be addressed and the best advice I can give you is to be familiar with it's faults so that you can work around it based on the tools that are available to do a more reliable check such as the Number Verifier as an integrated verification plugin for Studio or external tools such as Xbench or Verifika that also have plugins and can work well with Studio.

    Paul Filkin | RWS Group

    ________________________
    Design your own training!

    You've done the courses and still need to go a little further, or still not clear? 
    Tell us what you need in our Community Solutions Hub

    emoji


    Generated Image Alt-Text
    [edited by: Trados AI at 4:30 PM (GMT 0) on 28 Feb 2024]
Reply
  • As for "Can I just ask if you are using the language specific check" I believe you are referring to a function of studio 2019 but we use the latest version of 2017. 

    Noted... good to clarify this in the intial post.

    In general I can't give you good answers for your questions for the following reasons:

    • number/measurement recognition in Studio needs work
    • we can see what is happenning wrt recognition in the source fairly easily (although not from your screenshots), but it's not so easy to understand what's happenning in the target.  For example:
      Screenshot of Trados Studio showing QA Checker 3.0 errors: repeated word found, number missing in target segment, and extra measurement in target segment.
      All of these segments have weird results (excuse the JP as I had to guess) but I'll use this last one to explain what I mean:
      - source clearly recognises three numbers
      - target sees two numbers and a measurement (92, 92 and 32 mm)
      - repeated word... Studio seems to smoking something here!

    I would normally suggest switching off the out of the box QA for this and use the Number Verifier as it does provide more granular control but then you lose the ability to use parts f the QA checker that are still useful.

    I'll share this with support, as usual, but there isn't a good workaround for this.  This part of Studio needs to be addressed and the best advice I can give you is to be familiar with it's faults so that you can work around it based on the tools that are available to do a more reliable check such as the Number Verifier as an integrated verification plugin for Studio or external tools such as Xbench or Verifika that also have plugins and can work well with Studio.

    Paul Filkin | RWS Group

    ________________________
    Design your own training!

    You've done the courses and still need to go a little further, or still not clear? 
    Tell us what you need in our Community Solutions Hub

    emoji


    Generated Image Alt-Text
    [edited by: Trados AI at 4:30 PM (GMT 0) on 28 Feb 2024]
Children
  • Paul

    Thank you for attempting to replicate this on your computer too. Seems you've found a similar error and recognize the issue and I appreciate you admitting that the function needs work. I'll take this as an answer and forward this to "ideas."
    Just to clarify, having number tokenization on or off makes no difference. Having 2 numbers in the source side and 3 numbers in the target side makes no difference. It seems that A) "Ignore numbers" is not working and B) the multiplication symbol [×] is not recognized as a character. Not the letter [x] which I assume would work actually... and yep. No error. It's the symbol that's the problem.

    92角×32厚⇒    92 × 92 × 32 mm ⇒
    92角×25厚       92 x 92 x 25 mm

    Trados Studio screenshot showing a 'Repeated word found' error in the 'Messages' panel for a segment with multiplication symbol.

    First example uses the multiplication symbol (gets error/note), and the next example uses the letter [x] and gets no error. 

    Anyway, enough of that. Thanks for your help. I'm trying to making the verification function useful for my team hence the pickiness. 

    emoji


    Generated Image Alt-Text
    [edited by: Trados AI at 4:30 PM (GMT 0) on 28 Feb 2024]