Individualizing a cognitive model of students’ memory in

Intelligent Tutoring Systems

Maria Virvou, Konstantinos Manos

Department of Informatics

University of Piraeus

Introduction

• In this study we will describe the student modelling module of an educational application.

• This module measures-simulates the way students learn and possibly forget by using principles of cognitive psychology concerning human memory

Student Model

the student model takes into account:

• How long it has been since the student has last seen a part of the theory

• How many times s/he has repeated it

• How well s/he has answered questions relating to it

Test Bed

• To test the generality of our approach and its effectiveness within an educational application we have incorporated it in a knowledge based authoring tool. The authoring tool is called Ed-Game Author (Virvou et al. 2002) and can generate ITSs that operate as educational games in many domains

Cognitive model (Ebbinghaus, 1998)

• t: is the time in minutes counting from one minute before the end of the learning

• b: the equivalent of the amount remembered from the first learning.

• c and k : two constants with the following calculated values: k = 1.84 and c = 1.25

kt

kb

c

)(log

*100

Learning Process

• Whenever a students encounters a new part of the theory, the date and time are stored in the system’s database (TeachDate).

• Whenever a student “uses” a part of the theory, the date and time of this action is also stored in the system’s database (LastAccessDate)

Retention Factor (RF)

• The Ebbinghaus’ model is generic. To personalize it we will use the Retention Factor.

• The RF is a base percentage of the how much of the fact a student actually remembers

• The RF is modified based on the student’s profile and progress during the test

Retention Percentage

• b is the result of Ebbinghaus’ power funtion if we set t=Now-TeachDate

• X is the Retention Percentage

• If we want to know how much of a fact (Retention Percentage) a student remembers at a specific moment then we use the following formula

RFb

X *100

%

Custom Retention Factor

To customize the Retention Factor of a student we will introduce two new factors:

• Memorize Ability Factor• Response Quality Factor

Memorize Ability Factor

Based on the student model we define the memorize ability factor, a constant number with the following values:

• 0 Very Weak Memory• 1 Weak Memory• 2 Moderate Memory• 3 Strong Memory• 4 Very Strong Memory

Memorize Ability Factor (2)

• The new range for the RF is from 90 (very weak memory) to 100 (very strong memory)

Memorise Ability Value

Retention Factor Modification

0 RF` = RF – 5

1 RF` = RF – 2

2 RF` = RF

3 RF` = RF + 2

4 RF` = RF + 5

Response Quality Factor

During the test, depending on the student’s answer we define the Response quality factor, a constant number with the following values:

• 0 No memory of the fact• 1 Incorrect response; the student was “close” to

the answer• 2 Correct response; the student hesitated• 3 Perfect Response

Response Quality Factor (2)

• At that point the RF is again modified as shown in the following table

RQ Modification

0 RF’ = X – 10, set TeachDate=Now

1 RF’ = X – 5, set TeachDate = Now

2 RF’ = RF + (MA + 1) * 3

3 RF’ = RF + (MA + 1) * 4

Response Quality Factor (3)

• When a student gives an answer, the modification of his/her Retention Factor depends on his/her Memorise Ability factor

• In the end of a “virtual lesson”, the final RF for each fact is calculated. If this result is above 70 then the student is assumed to have learnt the fact, else s/he needs to revise it.

Conclusions

• We have described the part of the student modelling process of an ITS authoring tool that keeps track of the students’ memory of facts that are taught to him/her

• For this reason we have adapted and incorporated principles of cognitive psychology into the system

• In this way the system may know when each individual student may need to revise each part of the theory being taught