Idea Delivered

Process the following segment in the background, while the translator is editing the current one

This would really speed up the process of moving from one segment to another. In Wordfast Classic, when the translator is working on segment 1, the application is already working on getting data for segment two in the background. It's really a good idea, because when your translated segment is ready, you can then easily move to another one without waiting for the results to be presented.

One of the must-have ideas, I believe!

Wojciech Matyszkiewicz

  • Dear Daniel,

    I'm happy that Look Ahead is now part of the Studio, since this was a feature I came up with on this page (I realise, though that you had been working on it for some time already). Anyway, some users on Proz.com forum have noticed the cons of the feature - exactly the ones you once mentioned here, commenting on my idea. Namely, what about immediate fuzzy matches? When looking ahead, Studio can't provide you with the translations you have just created.

    I think I have come up with an idea how to deal with this problem. Here's what I posted on Proz.com:

    <i>At that time, the idea seemed simply perfect. When moving to the next segment, all the results are there, ready for you and you don't have to wait several seconds, especially when you have several large TMs and MTs in the project.

    However, Daniel from SDL pointed out that they were considering implementing it, but there might be some problems that will emerge in the process.  In short, since Look Ahead processes the segments "ahead", it can't get your immediate fuzzy matches, namely those ones that are based on your translation done in the previous segment, because at the time it was processing the document, the translation did not exist.

    I think there might be a nifty way to deal with this problem in the future. Here's my idea.

    I suppose that when creating a new project, most of us use at least TWO translation memories. One Big Mama, which is used ONLY as reference (Concordance and TM lookups), and one initially empty TM, where new segments are stored. Now, if we were able to tell Studio, which TMs can be used for Look Ahead (Big Mama) and which NOT (the small project TM), then the problem would be solved.

    Studio can do lookups in small TMs very quickly, but it needs time for Big Mamas. Hence, enabling Look Ahead for the latter and disabling it for the former, would do the trick! In this way, we would kill two birds with one stone!

    What do you think?</i>

  • So perhaps, to avoid clashing with TM matches,  you could implement it for all resources but TM lookups? Or give the translator the option, which resources they want to look up in advance and which ones will be queried dynamically?

    Anyway, a great news!

  • Thanks - indeed has been considered for a long time. It's a bit tricky as the following segment(s) might get invalidated by the current segment (e.g. new fuzzy match created) so the dynamic lookup is the most precise. This means that we need to design this carefully but we are strongly considering it for future versions as indeed this would change the perception of performance quite significantly.