Re: [edict-jmdict] KanjiSimNet

[Jeroen Hoek <mail@*************>]

Dear Hans,

This looks really interesting, but it completely locks up my browser.
This is Firefox on Ubuntu (GNU/Linux). Something is taking up full CPU
the moment I open your link.

A grapheme distance matrix? Does this mean you are analysing glyphs
based purely on shape of the font you use or are you using a kanji
grapheme composition database such as Chise?

Kind regards,

Jeroen Hoek

2009/2/13 Hans-Jörg Bibiko <bibiko@eva.mpg.de>:
>
> Dear all,
>
> some while ago I kept busy with clustering Japanese kanji according to
> their graphical similarity of complexity (based on a grapheme distance
> matrix). Now I set up a tiny test script which visualise this
> similarity on-the-fly.
>
> If you have a moment I'd like to ask you to test it.
>
> http://bibiko.de/cgi-bin/getjs.cgi?q=
>
> * the chosen technology won't work with Internet Explorer; please use
> instead Firefox, Safari, etc.
> * click at kanji or type in an other one
> * drag the light grey circles
> * up to now it doesn't work for "simple" kanji like 一 二 etc.
>
> It is a beta version and I'd like to know whether it could be useful
> in general.
> A classical use case could be to look for a kanji via a similar one.
>
>
> Are there any comments?
>
>
> Cheers,
>
> --Hans