Korean has tons of particles, verb endings, etc. so there are a lot of words recognized as different on here that are really the same word just with different grammatical implications. For example, 학교에서, 학교에, 학교까지, 학교가, 학교를, and more are all just the word for school. I think with more accurate tokenization of those to just 학교 along with the same for other nouns, verbs, and such that the single-word dictionary lookup and flashcards for Korean would be much more useful.

I don't know how your backend is written but there is a pretty good Korean tokenizer written in Java with wrappers for various languages here: https://github.com/twitter/twitter-korean-text