HanziFinder

HanziFinder is a substructure search tool of Chinese characters.

We have a database of nearly 89K 字 from which we can extract a continuous stream of unlabeled characters.

We designate these characters positive or negative based upon the substructure in which we are seeking, i.e., 乃 as in 奶, 孕, etc., and then iteratively train the tool, through a neural network, to find more characters with the same substructure.

We are currently in the process of testing the tool.