I am not qualified to make accurate ids for Japanese characters–they’re too nuanced and complex for a non-native speaker. But I did get something out of this exercise, most prominently a good way to represent passive voice.
The general problems are interesting tho. For a single Japanese compound, like “seikatsu” (economy) I could make simpler replacements for the individual kanji or replace the compound with a single simple id. I did go through a book of kanji and make id replacements. I found it often difficult, mainly since a single kanji can have many disparate, diverse meanings. The same thing occurs in English though, with words like, for instance, “figure” and “get” with all of their meanings.