Now that sentence cards are out, Im considering going back to using hanzihero for words and characters since I started doing only characters toward the beginning of the year. But Im not sure how to go about doing that since the words it sends dont seem to be paired to me most recently learned characters and I already have sentence cards in other SRS systems corresponding to the older characters I learned without words.
How does it pick words? Does it account for me learning characters out of the the default order? Is it the oldest learned characters that havent had corresponding words added?
We have a documentation article that outlines some of it. To add some additional details, our algorithm roughly works like:
First, look at the “daily word limit” for the user. We don’t schedule any additional words once that is hit. It can be thought of the ratio between words and everything-else within the lesson queue.
Then look at the words “available” (i.e., the characters within them have become learned/familiar or skipped.).
Of those words, find the ones which have a character which has the LEAST number of already-learned words associated with it for the user. So if a word AB has character A and the user has learned 0 words that contain character A so far, then it would be prioritized above word BC if the user has learned some words for both characters B and C already.
Lastly, secondarily sort based on the HSK level of the words. So if a user has 0 words learned for A, but AB is HSK-1 and AC is HSK-2, AB will show up first.