Word selection algorithm visualization

 

The idea is quite simple. Whenever player gives a correct answer a word is marked as learned.

Probability of drawing any already learned word is much lower – 1 of 50 to be exact. We’re achieving it by drawing 50 words from the pull of total 1500 words. 49 are from unlearned group and 1 is from the learned group (if available). This gives a short list of 50 words from which we draw the final result.

Giving a wrong answer to a unlearned word doesn’t change it’s status. Giving a wrong answer to already learned word changes it back to unlearned one. No matter how many learned words there is in the big pull there will be always only one in the short list.

Saving learned words works within game/language combinations. Let’s say you’ve learned 20 words in polish playing game mode were you have to choose one image. Then you switch to other game without changing the language. Those 20 words you’ve learned won’t be learned in this new game/language combination. Thus you can play the easier mode at the beginning and switch to the harder one and check if you can for example type the words you already answered correctly before.