• State Suggested Answer
  • Replies 7 replies
  • Answers 1 answer
  • Subscribers 432 subscribers
  • Views 3771 views
  • Users 0 members are here
Options
Related

Uplift/Repair when no match in TM but with a match from termbase

Piotr Bienkowski

I was told it should work, but did not in my case.

There was no match from the TM, and I did not expect it, but there was a match from the termbase.

The only difference was that the source was all uppercase and the termbase term was lowercase.

It was a one-word segment. Yet the word was not inserted automatically.

The TM was upgraded and the search and repair settings were taken care of.

What conditions must be satisfied for it to work?
Rate translation
×
Suggest better translation
×
Moderator UI
×
  • Get AI Suggestion
Parents
  • 0
    Hi Piotr - I may be misunderstanding what you've described, but a one-word segment doesn't sound like a situation where fuzzy-match repair could usually be applied. If you have a 10-word segment in the TM, and (say) a 10-word segment in the document that differs by one word, then assuming various other information is available, match repair can alter the high-scoring fuzzy match accordingly. But with a one-word segment, it's an all-or-nothing affair; either there is a segment in the TM that consists of the same single word, or there isn't. If there is, it's a 100% match. If there isn't, then there's no 'fuzzy match' for match repair to work with. Does that correspond to the situation you encountered? If not, could you post more details of the example?
  • 0 in reply to Kevin Flanagan
    Hi Kevin,
    So, I understand there was nothing to repair. What about uplift? Shouldn't it pick the term from the termbase automatically? Unless I am misunderstanding the concepts behind uplift.
  • 0 in reply to Piotr Bienkowski
    Hi Piotr,

    Maybe misunderstanding a little. What do you expect upLIFT to do? It either offers a whole TU or a Fragment match for use in your segment but if it's not repairing a match then you have to enter it into the segment. If it was a whole TU, and because it was one word, then you should probably see a 100% match from the TM anyway. If it had found a fragment match then you would enter it using autosuggest... exactly the same as your termbase.

    Make sense?

    Regards

    Paul

    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

  • 0 in reply to Piotr Bienkowski
    Hi Piotr

    The upLIFT functionality in this release essentially covers two things:
    1. Leveraging fragment translations from your TM (and showing you those translations, to choose from)
    2. Repairing fuzzy matches.
    What I think you'd like to see falls into something I'd put in a third category:
    3. Building translations from scratch using termbase (or even TM fragment) matches.
    Studio doesn't currently do anything in the third category, mainly because more often than not, the resulting composite translation needs significant editing, and it works out quicker just to make the translation pieces available via AutoSuggest, so the translator can quickly create a correct complete translation. It's true, though, that if a document segment with no segment-level TM match has content that matches a single termbase entry (or perhaps a single TM fragment match), and no additional content, Studio could pre-populate the translation in the editor, with more chance of that being a correct translation. Given that it may not be, though, and given that it's so quick to select a single AutoSuggest proposal, I'm not sure that's definitely the best thing for Studio to do ...
Reply
  • 0 in reply to Piotr Bienkowski
    Hi Piotr

    The upLIFT functionality in this release essentially covers two things:
    1. Leveraging fragment translations from your TM (and showing you those translations, to choose from)
    2. Repairing fuzzy matches.
    What I think you'd like to see falls into something I'd put in a third category:
    3. Building translations from scratch using termbase (or even TM fragment) matches.
    Studio doesn't currently do anything in the third category, mainly because more often than not, the resulting composite translation needs significant editing, and it works out quicker just to make the translation pieces available via AutoSuggest, so the translator can quickly create a correct complete translation. It's true, though, that if a document segment with no segment-level TM match has content that matches a single termbase entry (or perhaps a single TM fragment match), and no additional content, Studio could pre-populate the translation in the editor, with more chance of that being a correct translation. Given that it may not be, though, and given that it's so quick to select a single AutoSuggest proposal, I'm not sure that's definitely the best thing for Studio to do ...
Children