• State Suggested Answer
  • Replies 18 replies
  • Answers 2 answers
  • Subscribers 432 subscribers
  • Views 5138 views
  • Users 0 members are here
Options
Related

With Studio 2021, machine-translated segments in TMX files are no longer marked as AT, but as 100%

Dennis Boffy

Hi all,

I have a client who sends me TMX files with every job containing machine-translations (created by the client's proprietary machine-translation engine) of all segments in the job. These segments have a CreationID of "MT!", which marks them as machine translations. Before starting work, I import these TMX files into the TM for the job. With Studio 2019, these segments would appear as AT (automated translation) in the editor, making it easy to see that they are machine translations that should be reviewed closely. The client's TM would also contain fuzzies and 100% matches which would be presented correctly as such.

After I upgraded to Studio 2021, the machine-translated segments are being presented as 100% matches, making it impossible to identify them and to distinguish them from real 100% matches during translation work.

As far as I remember, I never changed any settings in Studio 2019 to get it to recognize machine-translated segments and mark them as AT, so I don't know what I can do to get the same behavior in Studio 2021, if anything. Can anyone help?

Thanks a lot in advance!

Best,

Dennis
Rate translation
×
Suggest better translation
×
Moderator UI
×
  • Get AI Suggestion
Parents
  • 0

    I have a feeling this fell into the category of "good bugs/bad bugs" where this was a bug that you actually found useful... so a good bug.  Most users do not want this behaviour because normally anything in your TM has been corrected and approved by you and the fact it originated as MT is now irrelevant.  The entry should be a correct translation unit. I can recall this behaviour, and it was also something I didn't like ether.

    I think that if you receive TMX files that are actually MT then the better approach is to import them into a new TM and then simply penalise the TM.  This way you'll never get a 100% match and will always review closely.  You could also add a field into your TM stating the origin and you would also get a good visual cue in the TM results window as you work.

    Paul Filkin | RWS

    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

  • 0 in reply to Paul Filkin

    Has this behaviour also been introduced in the latest Studio 2019 update (I have now SR2 15.2.8.3007)? It seems I have this situation just now after updating a few days ago.

    I pretranslated with setting to confirm 100% matches. Does this add those matches also to the updatable Main TM?
    I'm asking because I later disabled the TM with the "Created By MT!" segments but then they still showed as 100% matches in the Translation Results - but apparently coming from me from my Main TM. So, in effect this made the situation even worse.
  • 0 in reply to Angela Sigee

    Hi Angela,

    I just updated my Studio 2019 to the same version as you, and MT TUs are now being shown as 100% matches in Studio 2019 as well... Very unfortunate that SDL introduces behavior in Studio 2019 which they have already said that they will reverse in Studio 2021.

    I can't say if your 100% matches are coming from your main TM - perhaps they are just present in the file because you did a pretranslation and Studio considers MT TUs as 100% matches? Either way, it has made Studio 2019 equally difficult to use with MT TUs as 2021, so let's hope that SDL very quickly get the update to 2021 out that will remove this behavior (and perhaps also revert to the old behavior in Studio 2019).
  • 0 in reply to Dennis Boffy

    Apologies both - this should not have happened but did (too long to explain here). As a workaround, you should be able to set a filter for all TUs except creation user "MT!" so you'll get a penalty of the "MT!" TUs at least.  We are close to a fix for Studio 2021 and I actually have a question around that. In our current pre-beta version, the FIRST time you see an AT match (Created by MT!), it will be correctly shown as AT so you can review/change it. If you confirm it with no changes, then from then on it will show as an exact match from TM. Would that be OK? In the past (Studio 2017), it showed as "AT" for eternity. I am not sure that was correct as once you had a chance to review and confirm, even if unchanged, it should now be an exact match. Would you agree with that change, or should it stay "AT" for eternity? (This would IMHO be slightly inconsistent with what we do with alignment units where the first time they have a penalty but once you confirm they turn into proper TM matches. We were thinking to apply the same logic here). Can you advise? Thanks, Daniel

    Daniel Brockmann
    Team Trados @ RWS

  • 0 in reply to Daniel Brockmann

    Hi Daniel,

    Thanks for the update! And I agree with you that once an AT match has been confirmed, it should from then on show up as an exact match, not as an AT match anymore. Once it's been confirmed, it doesn't really matter anymore where the TU originated (as MT), the important thing is that it has been reviewed and accepted and so does not need the same level of scrutiny from then on as an MT TU, only the "normal" context-based check as other 100%/CM matches.
  • 0 in reply to Daniel Brockmann

    Yes, I think that would work (for both MT and alignment matches), for me at least. May be making it a setting. Communication is key here because it could seriously trip up translators.

    I wonder about my question whether pretranslation confirmations are written into the main TM. I mean I can test it but if you knew it would save me the effort.
Reply Children