I would like to do some simulations on progress using an SRS program (Anki, to be precise).
For this it would be useful to have some reasonable mathematical model of recall within the context of SRS-assisted memorization. For my immediate purposes, I would like to have a random function like
test(number_of_reps, time_since_last_seen, average_score)
to simulate the outcome of presenting a card that has been tested number_of_reps before, most recently time_since_last_seen ago, and having an average score so far of average_score. This random function returns a new score, which can be one of three values: 1 (wrong), 2 (correct with difficulty), 3 (correct easily), and the relative frequencies of these three values approximates the distribution one would get from performing the test with real human subjects.
Can anyone point me to some paper where I can derive this function from?
TIA!
For this it would be useful to have some reasonable mathematical model of recall within the context of SRS-assisted memorization. For my immediate purposes, I would like to have a random function like
test(number_of_reps, time_since_last_seen, average_score)
to simulate the outcome of presenting a card that has been tested number_of_reps before, most recently time_since_last_seen ago, and having an average score so far of average_score. This random function returns a new score, which can be one of three values: 1 (wrong), 2 (correct with difficulty), 3 (correct easily), and the relative frequencies of these three values approximates the distribution one would get from performing the test with real human subjects.
Can anyone point me to some paper where I can derive this function from?
TIA!
