We will introduce several basic terms for our model.
In the further text, we will use the term Fact as a generalization for a word, phrase, era, or any other one piece of simple information we want to learn.
Retrievability is a number which expresses the ability of a student to recollect a certain fact. We will express it in percents. If we say that a Fact has (in the very moment) RET=70%, it can be interpreted as follows:
- There is a 70% probability, that I will recall the Fact.
- From 100 such Facts, I recollect 70.
- If I evaluate the testing of Facts on a scale of 0-4 points, I’ll get 3 points. Therefore I’ll make one small mistake or I am not completely positive.
- Average point evaluation in testing more facts will be 3 points.
Retrievability decreases over time. In the language of the newly established terms, we say that we are forgetting. The Stability is a parameter which says how quickly the Retrievability disappears. The Stability can be increased by a suitable repetition of the Fact. That is the goal of learning.
Optimal Forgetting Index (OFI)
How do I remember the Fact? I have to remind myself of it before I forget. Using the implemented definition of RET, one can say: I will remind myself of the fact before its RET drops under a certain level. Let’s call the RET volume we admit to losing the Optimal Forgetting Index (OFI). Knowing the dependency of the RET on time, I can calculate from the appropriate OFI the next time I need to repeat the Fact This is the basic idea for the generalization of the simple advice “review after 1, 2, 4… days”. Let’s suggest some calculations.