PREDISPOSITION-BASED INTELLIGENT TUTORING SYSTEM

Adaptive User Profiling in Human-Computer Interaction

Andrzej Niesler and Gracja Wydmuch

Institute of Business Informatics, Wrocław University of Economics

ul. Komandorska 118/120, 54-345 Wrocław, Poland

Keywords: Predisposition-based teaching, intelligent tutoring system, human-computer interaction (HCI).

Abstract: The aim of the article is to present the conception of the new approach to human-computer interaction in

intelligent tutoring systems. Teaching supported by technology is one of the most important issues these

days. The crucial problem is how to learn effectively and how to substitute the student-tutor contact.

However, traditional education also does not provide individual, predisposition-oriented teaching programs.

Many of the people all over the world choose electronic way of studying. Therefore, it is crucial to evolve

methodology of user-adapted solutions. In this paper we present the learning predisposition-based tutoring

in intelligent instruction system, with individual user profiling, based on psychological conditions.

1 INTRODUCTION

Following the old Chinese proverb “Tell me, and I

forget, show me, and I remember, involve me, and I

understand”, one can make a clear conclusion about

the importance of the appropriate education process.

Traditional lectures are noticeably less effective than

exemplifying and drilling, because hearing alone is

not the strongest of the human senses. For efficient

learning, all of the senses need to be involved.

From the learner’s perspective, the process of

learning can be perceived as the combination of both

cognitive and psychical processes, concerned with

the assimilation of information. All these processes

encompass acquirement and improvement of skills

and knowledge, but also attitudes and behaviours.

In this paper we define tutoring as the activity of

providing knowledge and coordinating the process

of learning. We focus on the category of education,

as the instruction process supported by technology.

In our opinion, there is a significant issue with

efficiency in learning systems. The commonly used

mechanism of repetition provides means only for

memorizing some sort of information. What is more

important, however, is to understand the educational

material thoroughly and absorb it persistently. Most

of the tutoring systems available currently on the

market are focused on the cognition process, leaving

the psychological aspects uncovered. Combining the

two facets together may yield better results.

Common pedagogical methods used in electronic

teaching are dedicated mainly to the general group

of students, not to the individuals. The benefits from

the usage of instruction programs are therefore more

like in the traditional high-numbered class groups. It

means that the tutor has little time for each student

and cannot deliver the required level of attention.

The technology-supported tutoring system ought

to provide an instruction program for each of users

individually, adapting in a dynamic way to his or her

particular needs and preferences The adjustment to

such kind of factors is usually realised through the

use of customised interface themes, individual visual

arrangements, flexible course time scheduling, and

so on. Nevertheless, despite the increase in comfort,

it does not make the learning process more effective.

The differences in human nature, character and

individual predisposition for learning imply what

sort of teaching method should be applied towards a

particular human being. Leveraging the adaptation to

user’s needs, the predisposition-oriented approach

can be more effective than the traditional tutoring.

The article discusses the conception of a new

approach to intelligent tutoring systems, based on

user profiling according to their individual learning

predispositions. We describe a universal method for

user profiling, in spite of the domain of tutoring

program or performing institution. We focus on the

category of education as the technology-supported

instruction process from application perspective.

435

Niesler A. and Wydmuch G. (2008).

PREDISPOSITION-BASED INTELLIGENT TUTORING SYSTEM - Adaptive User Proﬁling in Human-Computer Interaction.

In Proceedings of the Fourth International Conference on Web Information Systems and Technologies, pages 435-440

DOI: 10.5220/0001532504350440

 SciTePress

2 INDIVIDUAL LEARNING

PREDISPOSITIONS

The conception of concentrating on the student’s

individual learning predispositions derives from the

psychological and pedagogical premises. In the

traditional teaching, the tutor’s acquaintance of a

student plays a very important role, as it is used for

choosing the best teaching method, leveraging his or

her skills and learning predispositions.

The tutoring model based on the individual user

predispositions can provide a real-time adjustments

in the general teaching program (Warren, 2005). The

advantage of such an approach is mainly in

operating with the consideration of the current user’s

needs. Therefore, a very significant increase in the

course overall effectiveness can be achieved.

2.1 Learning Determinants

There are three main determinants that can affect the

learning process. Those are the following:

 Memory,

 Understanding,

 Content association.

Memory concerns all of the activities that require to

keep in mind data and simple information. Next,

understanding represents cognitive process based on

the comparison between oneself interpretation of an

information and its objective meaning or intention of

the sender. Content association can be considered as

the ability to compare the information meaning, and

associate it in logical, causal-result relationship.

Figure 1 represents the mutual relations between

those three main factors in learning environment.

Figure 1: Memory, understanding and content association

connections schema in learning environment.

There are three standard elements of information

system (IS), as the domain of the learning process:

 Data,

 Information,

 Context of information.

Information is a form of structuralized data set.

Context of information may be provided by a user,

group, or community. Double-sided arrows between

information and its context represent the continuous

process of creating and using context to generate

new information. Process of enriching information

with context gives the base for creating knowledge.

We use ‘Memory’ for all three IS elements: data,

information (with participation of ‘Understanding’

process) and context (involving of ‘Understanding’

and ‘Content association’ processes).

Those three determinants can be activated in a

various ways by different individuals. It all depends

on their immanent predispositions for learning.

2.2 Basic Predispositions for Learning

In the history of psychology science, there were two

standpoints. One was gathering the followers of the

theory, that the entire human is similar and has

practically unlimited potential for development. The

other group claimed, that individual differences

between people are determined by the biological

conditions, which should be treated as guidelines for

upbringing and personality moulding. It implies the

necessity of creating the society structure that would

enable activities consistent with individual’s innate

abilities and predispositions (Anastasi, 1958).

Gerald M. Edelman (1972 Nobel Prize laureate)

proved the theory, that there is no identical brain

system, determined by different neural connection. It

leads to conclusion that the brain of each individual

is as unique as fingerprint. So are the predispositions

for learning and the corresponding user profile.

The essential role in researches over individual

differences plays the problem of reality perception.

It concerns the elementary experience processes,

such as: impressions, sensations and imaginations.

In practice we meet number of people whose mental

imagery is visual rather than auditory, or conversely.

The other factor is the process of memorizing. It

includes the following types of memory: semantic,

logical, visual, mechanical, associative, lexical, and

many others. The typology of memory depends on

the predispositions. It gives a variety of possibilities

for using memory in creating instruction programs.

Simultaneously, it brings the problem into teaching

methods’ effectiveness, because there are too many

inexactly defined categories of predispositions.

Learning environment

Memory

Understanding

Content association

Data

Information

Context

WEBIST 2008 - International Conference on Web Information Systems and Technologies

436

The solution appeared with intelligence category as

a factor responsible for the process of learning

(Guilford, 1967). We can analyse the Mensa

questions for measuring the intelligence quotient.

There are four categories of problems that concern

the perceptive, language, logical and numerical

predispositions. People always have one of the

abilities stronger than others. In that field we can

identify the predispositions, which can be useful in

the tutoring methods particularly.

2.3 Predisposition Matrix for Tutoring

Methods

Following the statement, that every person has the

individual suite of learning predispositions, we can

conclude that the tutoring program, as well as its

instruments, should be adapted to the needs of a

potential student. There could possibly be a problem

in defining unique instruction path for each student.

Therefore we need to find the way how to analyse

learning potential and how to create the model that

would help in defining tutoring methods.

First of all, as we now have the general learning

determinants, we should discern the instruments for

tutoring based on predispositions. Each instrument

represents the category of tutoring methods (table 1).

It results from learning determinants and represents

the predisposition aspect.

Table 1: Predisposition-based tutoring instruments.

Learning

determinants

Predisposition-

based tutoring

instruments

Instrument

method

high repetition

frequency

low repetition

short, intensive

training

Memory

time unit

long, extensive

training

analytics deduction

Understanding

synthetics induction

perceptiveness visualization

language

keywords, word

relations

logic

causal-result

relationship

Content

numbers

formula

representation

Directly from the ‘Memory’ determinant, we can

receive a frequency of the instruction program and a

time unit, designed to realise each exercise, task, or

problem. According to the ‘Understanding’ process

determinant, we have two ways of brain processing:

the analytic and the synthetic one. It corresponds to

the following methods: deductive and inductive.

For the content, we have specified instruments

based on the intelligence perception: perceptiveness,

language, logic and numbers. For each instrument a

different method of tutoring has been allocated:

visualisation, for those who are more perceptive,

keywords with the word relations, for the language

intelligence, the causal-result relationship problems,

for logic-guided people, and formula representation,

for the so-called mathematical brains.

For the usage of the previously mentioned three

main learning determinants, we have to allocate a

combination of all the instruments. Table 2 presents

the matrix of the predisposition-based methods in

aspect of memory, understanding, and content usage.

Table 2: Matrix of the predisposition-based methods for

tutoring programs.

MEMORY

UNDERSTANDING

high repetition

low repetition

short, intensive

training

long, extensive

training

visualization

keywords,

word relations

causal-result

relationship

CONTENT

formula

representation

deduction

induction

Because there are 4 instrument methods for the

‘Memory’ determinant, 2 for the ‘Understanding’,

and 4 for the ‘Content’, it gives 32 complex methods

for tutoring in total. This means that e.g. for one

individual there can be prepared a tutoring program,

specified by the following set of methods: intensive

training with high repetition of instructions, based

on deductive thinking with causal-result relationship

presentation of information.

3 USER PROFILING IN HCI

TUTORING

Human-computer interaction (HCI) is nowadays one

of the biggest issues in the IT development (Preece,

2002). In this paper, we perceive it mainly from the

PREDISPOSITION-BASED INTELLIGENT TUTORING SYSTEM - Adaptive User Profiling in Human-Computer

Interaction

437

Figure 2: Conceptual model of HCI tutoring framework.

perspective of a tool, used for increasing the tutoring

efficiency through the adaptation to user’s needs.

Technology usage methodology is determined by

the quality of the tutoring process optimization. It

implies the need for adapting interaction to the user

profile (Chung, 2004). The tutoring process is led by

the individual program for each defined profile.

3.1 User Predisposition Profiling

The major problem is to generate the profile that

would be the foundation of the interaction process

with the computer, with respect to the individual

preferences of a particular user. In the case of our

intelligent tutoring system, we have to concentrate

mainly on the learning predispositions.

User profiling is used for creating the individual

teaching environment. It consists in identifying user

preferences, such as the ability for learning. After

diagnosing of the user type, all suitable instruction

methods should be applied in the system. The data

computation helps with recording user parameters

and defining the suitable tutoring program. For the

communication between the user and the system, we

have the category of HCI framework.

3.2 HCI Tutoring Framework

The HCI framework is dedicated for tutoring. The

main element is the human constituent, i.e. the user.

Therefore, the HCI design should be user-centred.

The user executes the learning process by himself.

The computer part of the interaction process gives

the instruction set and invokes human reaction. HCI

tutoring framework is settled on intelligent system,

which plays the role of a teacher, so it has to respond

abreast of the student needs and behaviours. Figure 2

presents the HCI framework for tutoring.

On the computer side, we can see the two system

layers: representation, concerning data, information

and knowledge procedures used during computation,

and application layer, which includes interface for

communication with user and the applied instruction

program. On human side, we have the user and his

learning abilities. The processes on the left represent

learning determinants: memorizing, understanding

and association of absorbed information.

We can observe that the application layer is the

common field for the human and computer activity,

including interface, tutoring programs, and activated

user learning. The invoked process of knowledge

acquisition concerns two processes of the human

brain: cognition and perception.

There appears to be the necessity of implying the

concept of the individual predispositions, based on

the quality of executing processes of cognition and

perception, such as memorizing, understanding and

associating.

3.3 Adaptation of HCI Tutoring in

User Profiling

Adaptation consists in adjusting HCI tutoring to the

user profile, accordingly to the individual learning

predispositions. It means that there are significant

differences between each profile on the application

and the interface level.

Understanding

Association

Memorizing

USER

Cognition

Perception

Application layer

Data

Information

Context

INTERFACE

Tutoring

program

Learning

process

Representation layer

COMPUTATION

HUMAN

COMPUTER

WEBIST 2008 - International Conference on Web Information Systems and Technologies

438

Figure3: Model of intelligent predisposition-based tutoring system.

For the tutoring process, we can use the method

matrix, adequate to each single profiling. For the

information presentation (content determinant) and

the proceeding method (inductive or deductive), we

have to follow the principles of the interface design

(Wright, 2005). The time matter for memorizing

depends mainly on the application execution.

The need of profiling for the individual user and

adapting the tutoring process for particular abilities

and preferences, induces the problem of identifying

each profile and applying it into the system.

4 ADAPTIVE USER PROFILING

IN INTELLIGENT TUTORING

SYSTEM

In the adaptive approach to the user profiling in the

tutoring systems, we concentrate on the aspect of the

flexibility towards student, especially by using their

learning predispositions to achieve best effectiveness

results. The interaction process between human and

computer using intelligent solutions helps with the

adjusting program to the user abilities, but without

the need of any psychological knowledge of their

individual predispositions.

4.1 Premises of Intelligent Tutoring

An intelligent system application supporting fully

the predisposition-based tutoring program comes up

with the following elements:

 Creating tutoring program,

 User profile modelling,

 Building tutoring methodological apparatus,

 Designing interface class.

Tutoring program is the process of instruction,

created accordingly to specified teaching methods.

The user profile is a complex characteristic of a

student, that is based on his/her individual learning

predispositions. Methodology apparatus is the tool

for instructing, adaptive to the user profile.

Interface is the class of the possible instruction

scenarios. Each of them depends on the user profile

and the dedicated tutoring program (Powers, 2006).

Intelligent system is based on the expertise due

to evaluating user potential and adapting the most

suitable training program.

4.2 Model of Intelligent

Predisposition-driven Tutoring

System

Model of an intelligent predisposition-based tutoring

system is presented on figure 3.

In the model we have two inference engines. One is

responsible for the expert subsystem, assigned to

profiling, and the other for the method importing.

Profiling expert module is dedicated to acquiring

predisposition pattern from the base and comparing

with user predisposition, received through enquiring

process. Inference engine generates the user profile,

which is then used for creating the tutoring program.

Tutoring method expert subsystem is dedicated

to instruction program applying. Inference engine

TUTORING METHOD

EXPERT MODULE

INFERENCE

ENGINE

USER

Innate

Predispositions

Predisposition

Base

ENQUIRY

BASE

INFERENCE

ENGINE

USER PROFILE

ADAPTIVE

INTERFACE

TUTORING

PROGRAM

Method Base

Instruction Base

PROFILING EXPERT MODULE

PREDISPOSITION-BASED INTELLIGENT TUTORING SYSTEM - Adaptive User Profiling in Human-Computer

Interaction

439

uses the method base and adapts collected matrix of

methods to previously received user profile.

Instruction base is the collection of information

that is designed for the tutoring program. Inference

engine transforms tutoring data into the teaching

program. Tutoring program is based on the suitable

methodology adjusted to user profile and generates

adaptive individual interface for the particular user.

4.3 Practical View on Adaptive User

Profiling in Tutoring Systems

Tutoring program ought to be designed according to

the object-oriented paradigm, as well as the base of

instructions. It gives the opportunity to implement

any data to the previously defined methods and to

present it in a suitable to the user profile form.

We can also use a dynamic, content-based web

environment for the predisposition-based tutoring.

Every student has his or her own profile, despite the

same material of knowledge to acquire.

There is also an interesting possibility of sharing

different user profiles in order to compare the results

and selected ways of learning. Different learning

predispositions allow for comparing various kind of

information presentations, that has been already

learned. It might be also a good issue for knowledge

sharing and exploring new areas of perception.

5 DISCUSSION

As the HCI concept provides a wide area of possible

solutions, that ensues from many various domains of

interests, we cannot treat it separately. For the best

adaptation to the user, the holistic perspective should

be considered with all of the HCI aspects.

Presented model does not include description of

the semiotic premises for engineering process, and

requires some sort of additional combining with the

defined predisposition determinants. Ethnographic

and language problems may appear while modelling

the predisposition base, and, therefore, should be

considered as a separate research issue.

The implementation difficulties for intelligent

tutoring system based on user’s predispositions may

be caused by insufficient psychological and social

expert background. The background constitutes the

foundation for the adaptive user profiling as much as

for technological advance and system integration.

6 CONCLUSIONS

Electronic teaching can achieve the advantage over

the traditional methods, mainly because of the

possibility to simultaneously use the different types

of media. Model of intelligent predisposition-based

tutoring system, presented in this paper, gives the

opportunity to provide individual and user-oriented

instructions with the support of the technology.

Psychological approach offers pedagogic ground,

which is the crutial element in the learning process,

and is being so often passed over by the electronic

teaching systems designers. The authors believe that

the proposed solution is universal and can be applied

in almost every kind of education scenarios.

Matrix of predisposition-based tutoring methods

is an illustration model, presenting our approach to

the problem. The matrix can be freely extended of

additional determinants or tutoring instruments. It

can give more complex method apparatus that might

give more accurate tutoring programs in the end.

REFERENCES

Anastasi, A., 1958. Differential Psychology. MacMillan

Company, New York.

Chung, E., Hong, J., Lin, J., Prabaker, M., Landay, J., Liu

A., 2004. Development and evaluation of emerging

design patterns for ubiquitous computing. In Proc.

DIS’04, Designing Interactive Systems. Boston, ACM

Press.

Guilford, J.P., 1967. The nature of human intelligence.

New York, McGraw-Hill.

Powers, D., 2006. Vision in HCI: Embodiment,

Multimodality and Information Capacity. In Proc.

HCSNet Workshop on the Use of Vision in Human-

Computer Interaction. Canberra, Australia.

Preece, J., Rogers, Y., Sharp, H., Eds., 2002. Interaction

Design: Beyond Human-Computer Interaction. John

Wiley and Sons, Inc.

Warren, I., 2005. Teaching Patterns and Software Design.

In Proc. ACE 2005, Seventh Australasian Computing

Education Conference. Newcastle, Australia.

Wright, T., Noble, J. and Marshall, S., 2005. Using a

System of Tutorials and Groups to Teach User

Interface Design. In Proc. ACE 2005, Seventh

Australasian Computing Education Conference.

Newcastle, Australia.

http://www.mensa.org.

WEBIST 2008 - International Conference on Web Information Systems and Technologies

440