I am not using machine translation, only normal TMs, but for example the following has a score of 95% even though it is AT, does anyone know why and is there a way to make these become 100% matches?
Needs a bit of history to explain this one (courtesy of Kevin in development). Originally, AT matches could only be produced for segments with a single token, where the token was Number, Measure or Uri. The score produced for Number was 100, but 95 for the others. This was probably because for anything other than Number, it's hard to be certain that the auto-generated translation is correct. When AT matches were extended to include other tokens this precaution was retained for the more ambitious generated translations. The 95% score for AT is hard-coded and dates back to at least 2012. It cannot be changed without development doing some work to either feel confident that 100% would now be appropriate some 9-years later, or create some new capability to make this user definable.
Hope that helps?
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
Thanks for the answer. At least I know why they are 95% now!
Thanks for the answer. At least I know why they are 95% now!
Translate
