[qhanzi] Changing from "simplified" to "all" Chinese characters

[Ben Bullock]

The advertisement for qhanzi.com currently says that it does only simplified characters. The site was originally something I dreamt up based on a new algorithm for character recognition after doing an online course at Coursera. It also morphed into the "ignore stroke order" part of kanji.sljfaq.org, which is based on the old qhanzi code. The original reason for doing simplified Chinese characters was just that I had the data for them. 

Recently I've spent a lot of time trying to improve the accuracy of qhanzi.com both for misdrawn and correctly drawn characters, as a test bed for improving kanji.sljfaq.org. This meant going through page upon page of user inputs, and I found that a lot of site users aren't searching for simplified Chinese but for either Japanese or non-simplified characters, for example the following:

This isn't either a common Japanese or simplified Chinese character, but experimentally I constructed recognition data for the character, and it worked correctly as seen in the picture above. Looking at the user patterns of the site, it seems to me that there is a demand for this kind of "general Chinese character" search, and I think it is possible to construct recognition data from the existing data, given enough time and effort. So what I'm planning to do with qhanzi.com is to turn it into a general Chinese character recognition system, covering at least the Unicode hanzi up to the 0xFFFF value. It will continue to recognise the simplified characters too.

[Xiang Li]

Have you ever thinking of making this into an app? I was hoping there is an app where I can draw what I want on my phone and see if there is a corresponding Chinese word for it. It would be a great tool for me to teach Chinese.