When learning complicated new Kanji I am sometimes concerned that maybe a group of radicals frequently repeats other characters and it makes sense to give that group an aggregate meaning.

For example, "turkey" and "feathers" appear very frequently, and many RTKers including myself have grouped them into the meaning of "mattress" (see the character 曜). The problem is that you often don't know in advance what is going to end up getting grouped later on, especially once you move past the scope of RTK1 + RTK3.

I wrote a script to attempt automatic primitive discovery by systematically looking at what radicals are in common between various characters. For example, if you were to input 贅 you would discover that 敖 is a common component of this character and many other characters, and could assign it a new meaning. I'm still messing around with the way Kanji are rated, so it's not perfect or anything.

You can check it out here: http://foosoft.net/cgi-bin/kanji.py

What do you guys think? Cool? Lame? Whatever?

[ファブリス]

It could be useful for creating RTK extensions

Out of curiosity did you use kradfile/radkfile for radical information?

PS: come on spill out the beans on Guild Wars 2 !! ^^

[Katsuo]

FooSoft Wrote:When learning complicated new Kanji I am sometimes concerned that maybe a group of radicals frequently repeats other characters and it makes sense to give that group an aggregate meaning.

I made some notes to help out:

1. Primitive combinations that are not given a name. here
2. Characters useful for learning or simplifying others found earlier in the text. here

Katsuo Wrote:1. Primitive combinations that are not given a name. here

Ironically, it seems the 敖 primitive FooSoft mentioned above isn't on that list, because while it appears in 9 JIS2 characters, it's in none of the RTK characters. In this thread, I've proposed the name "sandbox" for this primitive (its actual meaning has to do with playing freely, which I think fits the notion of "sandbox" in more ways than one).

I'm creating a suite of apps in PyQt for messing around with Kanji and reading Japanese (think rikaichan but better), and this is kind of a prototype for one of the components. As far as the data I used, it's this guy here: http://foosoft.net/cgi-bin/radicals.dat , it's actually from the same data that comes with rikaichan.


@ファブリス: Nice try