Love it! Especially the generated png summary.
One suggestion for “unsupervised”/“unlabelled” ML scenarios like this is to give 3 choices and let user pick the one most meaningful to them.
For example, my daughter’s name is Alba. It went with “white” as primary translation and gave 白 (Bai2) but felt 小白 was a little too on the nose lol.
I changed the input to be “Dawn” (another meaning of the name) and got a somewhat better result of 晨.
Might be nice to provide an extra descriptor word too.
Still very cool and wish I had this when I lived in China. I went with the transliteration of my name: 安德斯 (Anders). Regret that I didn’t get one more meaningful!
Two nits: 1) I want to see the history of names I’ve previously generated 2) in the png, the headings on each side are different but I imagine the content is the same just a different language? I.e. “Character meanings” vs “character analysis”