My EN-JA machine translation software has a feature called "glossary learning."
When there are two or more glossary matches, the one that is selected is automatically applied first in the following segments (regardless of whether that glossary entry is the first entry or not).
Learn automatically as you translate.
Interesting, isn't it?
Sounds kind of like that thing that Hans is always talking about which I do not understand. He will know what I mean, although none of you probably will. Hell, neither do I.