Pretranslate AA overriding MT-filled segments and its associated enhancement CRQ-3937 still open

Hi Team,

The Pretranslate AA has couple of bugs to be fixed.

1) We notice that the Pretranslate AA is overriding MT translated segments regardless the minimum threshold I set in the AA, for example, our MT is set to translate 75% and any lower matches, but using this Pretranslate AA after the MT process in the workflow will override/strip off everything.

2) We expect this AA only pre translate given range of matches except ICE and 100% matches (those matches have been auto populated by the Segment Asset AA), so this AA should only for example pre translate 0-99% or 76-99% as such.

3) I filed an another bug long time ago which it has been become an enhancement CRQ-3937 since then, what is the status of this?

Are all these get fixed in 11.3?

Parents
  • Hi Anson

    thanks for the questions. The Pretranslate Auto Action does indeed operate on segments that are not 100% or ICE, and its intended use is to be paced early in the workflow after segmentation. The AA's description says: "This can be used to automatically provide high quality fuzzy matches in the target." And the status values you can set are either None, or Pending Review to indicate that it should operate before any human steps.

    Regarding the overwriting MTed segments, have you tried reversing the order of the auto actions in your workflow, so that Pretranslate provides any fuzzy matches from 76-99%, and then anything that is 75% or below goes through MT ? That seems to be a more logical flow.

    Thanks
    Ray
Reply
  • Hi Anson

    thanks for the questions. The Pretranslate Auto Action does indeed operate on segments that are not 100% or ICE, and its intended use is to be paced early in the workflow after segmentation. The AA's description says: "This can be used to automatically provide high quality fuzzy matches in the target." And the status values you can set are either None, or Pending Review to indicate that it should operate before any human steps.

    Regarding the overwriting MTed segments, have you tried reversing the order of the auto actions in your workflow, so that Pretranslate provides any fuzzy matches from 76-99%, and then anything that is 75% or below goes through MT ? That seems to be a more logical flow.

    Thanks
    Ray
Children
No Data