Fig. 1.

Examples of (user, item) pairs whose corresponding reviewing times have high likelihood under MEMORIZE (Top), threshold-based reviewing schedule (Middle), and uniform reviewing schedule (Bottom). In every plot, each candlestick corresponds to a reviewing event with a green circle (red cross) if the recall was successful (unsuccessful), and time $t=0$ corresponds to the first time the user is exposed to the item in our dataset, which may or may not correspond with the first reviewing event. The pairs whose reviewing times follow more closely MEMORIZE or the threshold-based schedule tend to increase the time interval between reviews every time a recall is successful while, in contrast, the uniform reviewing schedule does not. MEMORIZE tends to space the reviews more than the threshold-based schedule, achieving the same recall pattern with less effort.