search

keeleinstituut / tv-tolkevarav

Tõlkevärav (Translation Hub)
1 stars 0 forks source link

Volume analyse based on translation memories doesn't correspond to reality #719

Open MariusJulius opened 4 months ago

MariusJulius commented 4 months ago

From sample project Order [JUM-2023-12-VT-21], it can be seen that the analysis does not take memory matches into account at all. As if the entire text should be translated from scratch, although when looking into XLIFF, there are sentences with 90% and 100% memory overlap.

Nele tried to create a project again with the same file from yesterday to do the analysis herself. But the project Order [JUM-2024-02-T-34], which was created yesterday, still gives me the message 'File generation is in progress', and when I first went to the page today, there was also a pop-up window 'Error'.

That there is some stuff with larger source files, that we can't generate XLIFFs, but Mari herself was able to create XLIFFs from the same source file, under her own project: Order [JUM-2023-12-VT-21]. Something was different in the system then.

In other words, we need to check

  1. the ability to generate XLIFF of large, voluminous source files
  2. translation memory analysis (whether the system takes memory matches into account in the calculation)

Äriseadustik-01.09.2023.xml.zip

kadmit commented 1 week ago

Blocked by: https://github.com/keeleinstituut/tv-tolkevarav/issues/793

MariusJulius commented 11 hours ago

@NeleKo can you make a new order and validate the issue?