[peterburkimsher]

If anybody is interested in a Kanji dataset for improved input methods, please check out the 52,835 characters I gathered.

https://blog.usejournal.com/making-of-a-chinese-characters-d...

I made my own input method for Chinese Hanzi, which decomposes the characters and lets me find characters based on their IDS codes. It also predicts words both forwards and backwards (in case you don't know the first character, but do know the second).

https://pingtype.github.io