Post-translation improvement issues

prev Conflict resolution thinking on the trade-off between translation quality and speed

from 2023/12/08

• Post-translation improvement issues
  • Approach to correcting the original text -> not always getting the expected translation
  • Approach to fixing the English text -> the question is what to do when the original text is updated.
  • Approach to fixing the translation system -> difficult
  • I guess I'll end up fixing the English text and not worrying about updating the original.
    • That's the thing with automatic updates.
    • let's see
    • You're forking when it comes to the manually fixed English text.

from 2023/12/10

• Post-translation improvement issues
• Two types of improvements
  • 
  • The type of improvement that makes what is not there become what it is.
    • This can be achieved by machine translating all Japanese articles.
  • The type of improvements that make some things better.
    • Humans need to start a cycle of improvement.
• Gradually spread through links from key intersections in the network of knowledge
  • 
• These two actions are called the same "improvement" but are different
  • So don't think that one system can achieve both goals.
  • If not designed well, one side A will interfere with the other side B.
    • The first step tends to be to start from A, which has a clear way of doing things, but as a result, B, which does not have a clear way of doing things, is hindered by the existence of A and has to be verbalized, which makes it even more difficult to do things.
• 
  • A is for Automatic, H is for Human
  • The previous system only considered the "Japanese to English A" part.
    • As a result, when I read and found something I liked, I had the hesitation of thinking, "If I correct this, it will be overwritten when the original Japanese is updated," and this hindered efficient intellectual production.
• Given all this, the question arises.
  • I'm tracing the design of , which is now "convert to Markdown, then translate and publish in Quartz," but does that fit my purpose?
  • I don't think I did.
    • English H reinterpreted as "human output" rather than Japanese downstream
      • Translation only assists in the "creation of English H
    • When English H is created, a fork has occurred and should be connected to Japanese by a two-way link.
    • 

from /villagepump/post-translation-improvement-issue#6575d3c1aff09e0000b7fdf6

This page is auto-translated from /nishio/翻訳後の改善の問題 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.