Multiradical Kanji Dictionary for your Keitai

As part of an elaborate not-getting-laid-at-all-cost strategy, I spent the best of my Friday night hacking at home on a whim, bravely ignoring 1am drunken phone calls from a lonely ex, I didn’t stop until I basically had a working prototype.

And thus here you go:
Dr Dave’s Keitai Kanji Multiradical Dictionary!

Of course, you can use this dictionary from any browser, but it has been made especially compact, so as to offer convenient browsing on a small keitai screen.

Why bother making yet another multiradical dictionary when Jim Breen (and many others, most likely) already offers a very decent one on his site?

Two reasons:

  1. I wanted one that be easy to use from a keitai. Jim Breen’s is still a bit heavy to load and browse with a small screen.
  2. I wanted a smarter system for radical selection. All the systems I’ve seen so far let you choose your radicals from a checkbox list of all common radicals. Such a list can be quite long. This makes finding each radical quite tedious and particularly cumbersome on a keitai. Mine use a slightly different approach, that requires at least some knowledge of basic kanjis, but make it much faster then.

Instructions

Fairly obvious, really:

  • Screen 1: enter a string of kanjis. Can be any kanjis containing one of the radical you want to match or directly a radical. In practice, this means you should pick kanjis that look similar to the one you are trying to match… Say, you want to figure out [汾], you could enter [分] and [海]…
  • Screen 2: you will get a list of all radicals matching any of the kanjis entered previously (in our example, you’d get: [ハ], [刀], [母] and [汁]). Select the ones that belong to the kanji you are looking for (e.g. [ハ], [刀] and [汁]). Optionally, enter a number of stroke, with a margin of error (if you want to get any stroke count, do not change the ‘all’ value).
  • Screen 3 will give you a list of all kanjis (if any) containing all the radicals selected in the previous screen, ordered by frequency and stroke count (in our example, you’d get only the kanji you were initially looking for: [汾]). Along with the kanji, you are given stroke count and unicode value. Clicking on the kanji will do a word search in WWWJIC (translations). Clicking on the unicode value, will give you WWWJDIC’s Kanjidic entry (kanji pronunciation keys and data).

This script has been successfully tested with AU’s EZweb, but should work on any net-enabled keitai, please let me know if you encounter any problem. Suggestions and general comments most welcome.

Hope you’ll find it useful, I know I will!

Note: As usual, this project uses extensively the amazing amount of data gathered and made available by the EDRG on Jim Breen’s website.

10 comments

  1. Hi,

    Neat tool. I’m a big fan of Jim Breen’s tools, and I also have started using “rikai xul” for Firefox (I’m not sure if this is such a good idea, but I like it so far…)

    I was wondering if it’s possible to create a dictionary search engine for Firefox. Any hints?

  2. Nevin,

    Not sure what you mean by a “dictionary search engine”… there are lots of different ways Jim Breen’s dictionary can be used with bookmarks or even direct access with a string lookup.
    Although personally, when not using a keitai, I much prefer local applications (I use JEDict for Mac, based itself on WWWJDIC’s word DB).

  3. neato – how are you doing the lookups btw? did you convert the WWWJDIC into something useful like a mysql DB? its nasty parsing their huge text files.

    I’m working on a thing to lookup matching images from flickr based on the englihs – and see if there really is any match to the kanji.

    any chance you can share some source for the above?

  4. dc: Yup, that I did… and indeed their huge text files are quite nasty… luckily nothing to frighten a few php scripts…

    Hmn, not sure what you are trying to do. Will be glad to share any source I got, although trust me when I say it’s wholly unimpressive (mostly a few sql queries with the wwwjdic behind it). Are you trying to do character recognition? That was one of my long-term pipe-dream, obviously infinitely harder than the current script (and others I’ve made) who all rely on the availability of japanese in text format…

    Anyway, feel free to shoot me a mail if you think I can help you…

  5. tx for the reply. heres what i’m working on:
    http://jgram.org/kanjipic/

    it would be nicer to break a kanji down to its radicals.
    for more complex words/ideas, the image results are not really that relevant. try 感謝 for example.

    what i want to do is make a little “kanji ticker” app that can run while working. but throw up photos too so it is a bit more fun to watch. not sure if there will be much relation between the images and the kanji tho.

    another idea might be to just parse the henshall mnemonics, and pick out meaningful photos for the “concept radicals” words in there.

    if you could send me a link to the DB stuff, that would be great. i would prefer to avoid parsing JDic and encoding issues…

    also havent started on parsing the radicals…which file did you use to get this data btw? radkfile?

    tx!

  6. Dr Dave-
    This is really useful. I think the component lookup method is far underused…it is THE way to lookup for intermediate students. It’s interesting that you call the parts of the kanji “radicals”…usually dictionaries want there to be only one radical per kanji. If you don’t know which part is the radical, you might have to look up a few times, and that assumes that you are familiar with all the different forms or names of each radical. Or by stroke count, so you sort through the dozens of 3-stroke radicals, then sort through dozens of kanji with that radical, only to realize that you got the radical wrong. This “traditional” method seems so impractical compared to component (multiradical) lookup.

    For me the fastest kanji lookup is the “buhin kensaku” in the “kanjigen” in electronic dictionaries (also available on the web for a fee). For example, I want to look up 汾 like your example, I enter “mizu” and “bun”, and it finds 4 kanji. So I pick 汾 from the four and I’m there. It takes no time at all. This skips the radical selection screen in your script, which can be cryptic since the parts are not in the same scale or context as they are in the kanji, and mistakes in selection are possible.

    I wonder if it’s possible to make a script to enter “分” and “海” (or “bun” and “umi”) and return all kanji with all the permutations of at least one component from both. Then instead of having to select from the radicals (and possibly make a mistake) you just pick from 20 someodd kanji. For intermediate students using a PC browser, I think this would be the fastest.

    My real question is: I didn’t know that Kanjidic had component (multi-radical) information other than the “radical number”. Is this information in Kanjidic, or did you invent a clever way to get it?

    To change the subject, it would be nice if electronic dictionaries had romaji-kanji conversion like PCs or cell phones, so this kind of script would be possible. I like the convenience of using the cell phone, but access it kinda slow.

    To change the subject again, your script didn’t work on my AU A1013K Kyocera cell phone using EZWeb…the phone’s a little old though. In the meantime, it still comes in handy at the PC…

  7. hi Dr. Dave,

    You have done excellent job as multi-radical Kanjidic.!!!
    Very thanks..
    But recently kanjidic site seems to be down. Because when
    i search any kanji it shows error as “could not connect to DB”.
    When it will be up?
    I am waiting eagerly to use it.

Leave a Reply