I found two public databases that describe the composition of Chinese characters.

1 - CJK Decomposition Data

2 - Chinese Characters Decomposition on Wiki Commons

I can see how the formats they use are different, but I don't think they are incompatible. Are these two data sources related or completely independent? Is it known which one is of higher quality?

The reason I'm asking is because I'm working on a tool that makes it easier to look up characters. It allows you to search by radical anywhere in the character, rather than by just the primary radical used in dictionaries.

  • 1. They are related, there're some rules for character composition. 2. CJK consists of Chinese, Japanese, and Korean. So it would be a larger set. 3. Quality depends on your criteria. Mainland China, Hong Kong, Taiwan have different standards. So, this is really a professional question ... it's difficult. Maybe you can collect data from this site, there're many experts helping refining their sources -- I think it is the best Chinese online dictionary, currently.
    – Stan
    Commented Aug 4, 2013 at 18:09
  • Sjors, what are you using the data for? When you don't say explicitly, people on this site tend to assume you are using whatever you ask for to study Chinese, which I doubt is the case here. Commented Aug 8, 2013 at 16:41
  • @StumpyJoePete I updated my question to explain. Commented Aug 10, 2013 at 15:29
  • 1
    You should check out this question and its answers. I can't comment on the difference in quality, but there are several sources of such data, and a tool already exists that works like you want. Commented Aug 10, 2013 at 21:50
  • @StumpyJoePete thanks. I assume you're referring to Tatoeba? That is indeed similar to what I'm trying to do and they are using the Wikimedia data. Of course I stubbornly believe I can do even better :-) Commented Aug 11, 2013 at 8:26

3 Answers 3


I would say that the CJK Decomposition Data (your first link; henceforth CJKDD) is better of the two, for most purposes. Some time ago I used the character decomposition data from Wikimedia Commons, and there were quite a few popular characters with missing decompositions. Now I checked a few of these characters, and all of them had decompositions in CJKDD, so it seems more complete.

The Wikimedia database is better than before, but still lacks decompositions of some characters, such as 齒, 風 and 龠. Moreover, some decompositions are incorrect, e.g. 辰 is shown as a component of 脈. And in general, the Wikimedia database doesn't show components that aren't separate characters. CJKDD, conversely, contains all kinds of components, including the ones that don't have their own Unicode code points.

Wenlin with its Character Description Language (CDL) is another source of decomposition data that you may want to look at. Even CJKDD has inconsistent descriptions of some characters, and Wenlin seems a bit better in this regard. For example, even though 𨊠 is present in the database, it is not shown as a component of 範 (範 is decomposed as 範:d(⺮,38298), which is in turn decomposed as 38298:a(車,㔾). In Wenlin, 範 is correctly decomposed into 𥫗 and 𨊠. Wenlin uses Unicode Private Use Areas for encoding components that aren't separate characters.

On the other hand, Wenlin has its peculiarities and, for example, decomposes 齒 into 𣦊 and 凵. CJKDD splits 齒 into 止 and 𠚕, which seems much more natural. In the result, Wenlin doesn't see 𠚕 as a component of 齒. Similarly, 曲 a sub-sub-component of 曹 in CJKDD and Wikimedia, but not in Wenlin.

Another difference between the databases is that CJKDD decompositions are consistently graphical. For example, 條 is decomposed as 條:a(39752,条), which is in turn decomposed as 39752:wr(亻,㇑). In Wikimedia Commons, 條 is decomposed into 攸 and 木, which has more to do with the historical development of the character and its radical (which happens to be 木), but not with its modern form. Wenlin provides both the graphical decomposition (based on the CDL data), and the list of all the sub-components that may be associated with this character (in this case: 攸亻丨条夂木). Another example: because the CJKDD decompositions are purely graphical, the upper right part of 祭 is shown as 卩. Wenlin, more correctly, shows it as a graphical variant of 又.


I'm not an expert in CJK decomposition, but I can share some of my knowledge as a native Chinese.

Both decomposition are purely structural I would say. It is suitable for computer processing (input and recognition). We also use this way to look up a character in dictionary (99% of it actually, there are exceptions).

So in a broad sense they are compatible. But just as different Chinese dictionaries may put the same character under different radicals (部首 bùshǒu), there is no guarantee of consistency and mostly depends on the editor of the classification. Nevertheless these discrepancies should be treated as exceptions rather than norm.

Reference on structural decomposition: http://chinesenotes.com/chinese_fonts_structure.php

More on CJK characters

Chinese character decomposition as a study is much more than that. None of the above decomposition referred to 六書 liùshū, the six way of Chinese character formation. I highly recommend you look into it if you are interested in Chinese characters.

Japanese has its own fixed set of phonetic characters (仮名 kana) and Chinese characters (漢字 kanji). It uses the rules in 六書 liùshū (mostly 會意 huìyì) to generate many kanji that is only used in Japanese and only have Japanese articulation.

Korean characters (hangul) are actually shapes of consonants and vowels packed in a square so it is also a fixed set of phonetic characters.

p.s. maybe a bit off-topic, sorry :-P

  • 1
    The "K" in CJK doesn't [only] refer to hangul but [also] to the Korean usage of Chinese characters.
    – dda
    Commented Aug 12, 2013 at 5:49
  • I have in front of me the New Practical Chinese Reader (the German translation actually). It list many decomposition and since this is apparently a very well-researched book I wonder what its scholarly reference regarding "C" decompositions may have been.
    – Drux
    Commented Oct 11, 2013 at 17:41

There are the Chinese characters and the Chinese radicals in the Chinese language. A set of radicals form a Chinese character.

Quite often a radical itself can be a Chinese character as well.

The Chinese characters can be decomposed (or broken down, split) into radicals for the sake of making the Chinese characters learning process simple and understandable.

That is why Chinese character decomposition is a very effective tool for teaching, learning and memorizing the Chinese characters.

Please see below how Chinese character decomposition makes learning simple, indeed.

The Chinese characters 口 kǒu ‘mouth’ and 日 rì ‘sun’ both include the radical 冂 jiōng ‘down box’.

In addition, the Chinese character 口 kǒu ‘mouth’ also includes the radical 一 yī ‘one’. The Chinese character 日 rì ‘sun’ also includes the radical 二 èr ‘two’.

So, the overall decomposition of the both Chinese characters looks like following:

口 kǒu ‘mouth’ 冂 jiōng ‘down box’ 一 yī ‘one’

日 rì ‘sun’ 冂 jiōng ‘down box’ 二 èr ‘two’

Simple, isn’t it? Now we understand that any Chinese character can be decomposed into the Chinese radicals.

  • That's not entirely correct. Each character contains a single radical, and not multiple. It's used to categorize characters in dictionaries. Commented Jun 11, 2018 at 23:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.