"Before Interactive Editing" after using Trados Batch Anonymizer

Hi,

I translated some file with NTM. Then, I wanted to anonymize information about the NTM engine, so I used the "Trados Batch Anonymizer". However, segments translated with NTM can still be easily recognized. Please have a look:

Below, a segment translated from scrach by me:

Close-up of Trados Studio Translation Details showing Status: Translated, Origin: Interactive, Score: 0%.

Below, a segment translated using NTM engine:

Trados Studio Translation Details expanded view with Status: Translated, Origin: Neural Machine Translation, Provider: OpenAI provider for Trados, Score: 0%, and additional Before Interactive Editing details.

And below, a segment translated using NTM engine and anonymized using "Trados Batch Anonymizer":

Trados Studio Translation Details showing Status: Translated, Origin: Interactive, Score: 0%, with Before Interactive Editing details repeating the same information.

As you can see, there is this additional information "Before Interactive Editing Origin: Interactive Score:0%." It appear only next to NTM translated segments. How to get rid of it?



Generated Image Alt-Text
[edited by: RWS Community AI at 6:55 AM (GMT 1) on 15 Oct 2024]
emoji
  •  

    I don't think this means they were NMT.  It only means that they were edited manually before accepting the final translation.  If you want to disguise the fact NMT was used at all then use the anonymizer to provide the name of a TM:

    Anonymizer settings dialog with options to anonymize username, resource origins, and Language Weaver provider metadata. 'Removechange MT Provider' and 'Set the specific resources used' are checked.

    Then the results show this:

    Translation Details box showing Status as Draft, Origin as Interactive with a Score of 80%, and Before Interactive Editing details with Origin from Translation Memory, Provider as myTM, and Score of 80%.

    So now it just looks as though you corrected a low fuzzy match from your TM.  I can only see this if I do correct the NMT anyway so this seems reasonable.  I also tested using the SDLXLIFF Anonymizer and this takes a similar approach.

    If you'd like to be able to remove every trace of having done anything to the file in the previous origin you should search for this:

    <sdl:prev-origin.+</sdl:prev-origin>

    Replace with nothing... using a text editor that supports regular expressions.  I'd back up first in case you make a mistake but that will remove it and it'll look the same as your translated directly by you.

    You could also raise this as an enhancement request here:  RWS AppStore Ideas 

    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: RWS Community AI at 11:19 PM (GMT 0) on 23 Nov 2024]