At https://github.com/darius/spaced-out I tried to do something vaguely similar: from an aligned parallel corpus, automatically make a prioritized spaced-repetition deck for language learning. (I think I used Europarl.) So you get examples of the words in context, plus they're sorted with the most frequent ones first.

(There's also an SM2-based flashcard reviewer in Python. It's all very crude; I decided I didn't want to learn Swedish enough.)