LEARNING PERSONALIZATION

Design Solutions in an e-Learning System

Ileana Trandafir, Ana-Maria Borozan and Alexandru Balog

National Institute for R&D in Informatics - ICI Bucharest, Romania

Keywords: e-Learning, learning personalization, learning design, learner model, educational content.

Abstract: This article describes some theoretical and practical aspects regarding the process of personalizing services

and content for an e-learning environment, issued within the national research project “Innovative System

for Personalized and User-centered Learning with Application to Project Management (SinPers)” developed

by the National Institute for R&D in Informatics, the Academy of Economic Studies and the Project

Management Association Romania. This project proved that the learning personalization needs innovative

solutions for three main domains: the design of the teaching-learning process (actors roles, activities

structure and flow, events and conditions specification), the creation and maintenance of an individual

model for each learner (goal, preferences, knowledge level, learning results) and the structuring and

accessing mode of the educational digital content (based on domain and competences ontology, learning

objects, metadata).

1 INTRODUCTION

1.1 Lifelong Learning - New

Requirements for e-Learning

Systems

In an information based society the lifelong learning

becomes an essential process, sustained mostly by

information and communication technologies. The

lifelong learning is a new form of work; the use of

knowledge acquired in school is made at the

working place, and the professional activity is more

and more relying on intensive-knowledge. Learning

becomes inseparable from the working process of

adults. Similarly, the children need new educational

instruments and environments to help them educate

their desire to learn and create. Lifelong learning is

more than “adult education”; it covers and unifies all

phases: intuitive learner (at home), scholastic

learner (at school and university), and skilled

domain worker (at workplace) (Fisher, 2000).

Several basic principles of the learning theory

have been re-evaluated in the last decade, as result

of the new services offered by the information and

communication technologies, as well as because of

the lack of success of the existing e-learning

systems. More and more critics sustain that the

simple use of ICT as support of the existing learning

practices is insufficient; old frameworks, such as

instructionism, fixed curriculum, memorization, out-

of-context learning etc., are not changed by the

technology itself (Attwell, 2007; Dondi, 2007).

New computational environments are necessary

to support new education paradigms such as lifelong

learning, integration of working and learning,

learning on demand, real-life problems, self-directed

learning, and information contextualized to the task

at hand, intrinsic motivation and collaborative

learning. The fulfilment of each user’s individual

needs (expressed explicitly or implicitly) - learning

personalization, educational content re-usability on

large scale - content reusability, assurance of the

communication between e-learning systems as well

as with other human resources management systems

- systems interoperability are the main objectives of

researches in this domain.

1.2 Personalization - An Advanced

Approach in the e-Learning

Systems

The learners have different learning styles,

objectives and preferences, which lead to variances

of efficiency and effectiveness of the traditional e-

learning systems from individual to individual. The

364

Trandaﬁr I., Borozan A. and Balog A. (2008).

LEARNING PERSONALIZATION - Design Solutions in an e-Learning System.

In Proceedings of the Fourth International Conference on Web Information Systems and Technologies, pages 364-369

DOI: 10.5220/0001517403640369

 SciTePress

learning personalization becomes an advanced stage

in the e-learning systems evolution.

The study carried out within the framework of

the SinPers project

(www.ici.ro/sinpers/) shows that

the concept of personalization can be interpreted and

implemented in different ways and proves that

learning personalization needs new solutions for a

multitude of aspects, such as: the identification of

the profile, goals and user context; the formalization

of knowledge; the description of learner

competences and learning objectives; the evaluation

of learner’s skill level; feedback return in an

adequate manner.

Within the SinPers framework, the

personalization issue was solved by adopting

innovative solutions in three main domains (figure

1):

1. the modelling of teaching-learning process,

2. learner modelling and

3. digital content modelling.

Figure 1: SinPers - personalization pillars.

A useful support was the adoption of the IMS

standard, which offers a conceptual framework for

all three mentioned areas of expertise (IMS, 1999-

2007). This choice was based on a global evaluation

of the existing e-learning standards (e.g. SCORM,

IEEE, IMS).

2 THE MODELLING OF THE

TEACHING-LEARNING

PROCESS

The modelling of the teaching-learning process

implied the evaluation of several alternative

scenarios having as objective the personalization of

the content and services offered to the learner, that

lead to the creation of a composed scenario

including several steps:

1. the specification of the personal training

options (e.g. entire course, one module,

competence acquirement), personal data (e.g.

studies / qualifications, age, activity domain)

and personal preferences (e.g. learning style,

hardware-software support),

2. pre-assessment of the learner knowledge level

according to its options,

3. personalization of the unit of learning (course,

lesson, module etc.) based on the learner’s

options, knowledge level, profile and

preferences,

4. unit of learning completion, with specific sub-

phases for a computer assisted course,

5. final assessment and course close-up.

The modelling of the teaching-learning process was

based on the informational IMS model (IMS, 1999-

2007), and was next transposed in XML. The roles

of different actors, learning activities, support

activities and the training environments (learning

objects and services) were defined in accordance

with the proposed scenario.

Essential elements were to define the method,

properties and conditions, on which are based the

personalization mechanisms as well as to control the

process execution. According to the standard, each

learner’s personalization is made in several ways:

• Activities Tree Personalization - through

definition of the plays, acts, activity structures

and role-parts,

• Environment Tree Personalization - similar

with the activities tree,

• Educational Content Personalization (selecting

and sequencing of the learning objects).

Thus, the personalization was specified explicitly (by

defining the conditions that determine the

completion of an act or activity, the plays and the

acts components of the teaching process and the

role-part relations), and implicitly (by specifying the

teaching process workflow, e.g. in a sequence of

activities, where an activity can be accessed only if

the previous activity was completed successfully). In

this case, the status of activities sequence must be

updated for each user. Regardless of the explicit or

implicit personalization method, the basic element is

the learner dossier. The maintenance of the learner

dossier is completed directly by the user through the

actions that were run (the event of completion with

success of a learning activity), or by a different actor

(the trainer during the support activities, e.g. a test

validation).

A detailed description of the design steps for the

teaching-learning process is presented in (Trandafir,

2007).

Learner

modelling

Teaching-learning

process modelling

Content

modelling

Personalization

LEARNING PERSONALIZATION - Design Solutions in an e-Learning System

365

The learning design (LD) specifications are the

foundation for the SinPers system architecture; some

extra functions have been added: management of the

educational content and administrative services

(figure 2).

Figure 2: SinPers architecture.

3 LEARNER MODELLING

3.1 Levels of Personalization

The personalization based on the learner model can

be achieved from five different angles or five

distinct levels, from simple to complex, as follows

(Martinez, 2000):

1. Name-recognized Personalization - is the

easiest solution and consists of the simple

acknowledgment of learners as individuals (e.g.

the learner name appears in the upper part of the

screen or previous activities or accomplishments

are marked);

2. Self-described Personalization - allows the

learners to describe preferences and common

attributes, the initial cognitive status or existing

skills, preferences, or past experiences (using

questionnaires, surveys, registration forms,

comments etc.);

3. Segmented Personalization - uses demographic,

geographic, psychological or other criterias to

group or segment the potential learners into

smaller, identifiable and manageable groups, for

personalization purposes;

4. Cognitive-based Personalization - uses

information about individual learning preferences

or styles, from a cognitive perspective, in order to

provide educational content in accordance with

these attributes of each learner. This

personalization type is more complex that the

previous ones and needs to handle more learner

attributes at each interaction with the system, by

collecting data, monitoring learning activities,

comparison with other learners behaviour and

predicting what the user would like to do or see

next;

5. Whole-person Personalization - assumes

profound understanding of the psychological

factors with major impact on the behavioural

differences in the teaching-learning process (more

profound than based on the cognitive profile). It

requires success in predicting and delivering the

necessary content, so that the learner can achieve

its objectives and - this is more important - to

improve its ability to learn and develop a personal

relationship with the online system. This

approach implies the consideration of multiple

emotional aspects, feelings, intentions that

substantially influence the learner’s behaviour

and evolution. As any individual, the system

learns as well by collecting data, tracking

learner’s progress, and comparing responses with

the correct ones in order to improve the responses

progressively. Therefore, it becomes more precise

over time. This is the most sophisticated

personalization form and requires real time

personalization in order to modify the responses

provided to the learner based on a dynamic

learner model that is changing throughout the

learning experience (as a teacher in class).

The learner model defined in the SinPers project

implements the personalization levels presented in

the table no. 1.

As emphasized in the this table, the learner

model is set up progressively, starting with the data

entered at user’s enrolment and continuing with the

specification of the objectives and preferences, the

assessment of the initial cognitive status and

learning styles (based on interactive pre-assessment

tests) and results tracing.

3.2 The Data Structure of the Learner

Model

At first the static and dynamic properties of the

learner within one unit of learning were

differentiated, by defining two distinct entities:

Educational

Content

Management

Educational

Content

Actors

Models

Unit of

Learning

Execution

Unit of

Learning

Personalization

Administra-

tive

Services

WEBIST 2008 - International Conference on Web Information Systems and Technologies

366

Table 1: The relationships between the personalization

levels and data model.

Personalization

level

Data model of

learner

Data gathering /

user interface

Name-recognized

personalization

Personal

identification

data

Learner

enrolment and

registration /

online registration

form

Qualifications,

Certifications,

Licences

Learner

enrolment and

registration /

online registration

form

Course objective Selection of the

menu option /

interactive

dialogue

2 Self-described

personalization

Pre-assessment

mark

Pre-assessment

test / interactive

dialogue

Age, Activity

domain, Studies,

Function,

Competences

Learner

enrolment and

registration /

online registration

form

3 Segmented

personalization

Pre-assessment

mark

Pre-assessment

test / interactive

dialogue

Language,

Format,

Technological

support,

Difficulty level

Learner

enrolment and

registration /

online registration

form

Unit of learning

personalization /

interactive

dialogue

4 Cognitive-based

personalization

Learning style Test to identify

the learning style

/ interactive

dialogue

5 Whole-person

personalization

Learner portfolio

(results, grades

obtained for each

activity)

Unit of learning

management /

tracing the learner

progress and

reporting

component

• The Learner Profile - containing the personal

properties set; these properties have an

invariant character during the unit of learning

execution and can be updated only at the

completion of the unit of learning,

• The Learner Portfolio - containing the

information set regarding the learner activities

and results during the unit of learning,

respectively recording and managing the

learner’s history of the training process, the

scopes and achievements / obtained knowledge.

According to the IMS recommendations, seven

segments have been used to define the personal

properties: identification, goal, qualification-

certification-licence, competency, accessibility,

affiliation, security key. Additional customisation

was performed, with respect to the standard (fig. 3).

Figure 3: Learner data model in SinPers.

The planned and achieved activities and partial or

final results (transcript) are the basic elements of the

learner portfolio.

Detailing the accessibility information as well as

the modality to create specific vocabularies for these

information categories according to course domain

and users categories (e.g. activity domains,

competences, course scope, security levels) are

original elements of this project.

The accessibility holds the most important

information needed to perform the personalization of

LEARNING PERSONALIZATION - Design Solutions in an e-Learning System

367

the unit of learning, respectively learner preferences

regarding: teaching language, educational objects

format, the technological support used (operating

system, browser), difficulty level (very easy, easy,

average, difficult, very difficult), as well as the

learning style (active/reflexive, sensorial/intuitive,

visual/verbal, inductive/deductive) declared or

established through testing.

The learning styles have been identified based on

recent research studies that analyse the basic two

steps of the learning process: collecting and

processing information, concluding in the most cases

upon the four styles mentioned above.

4 CONTENT MODELLING

In order to meet the personalization options, the

content of SinPers “project management” course is

structured as a collection of distinct learning objects

(LO). Their reuse in different contexts and

(re)sequencing in different learning paths requires

the adoption and definition of two essential

elements:

• domain ontology (the structure of concepts

and the relationships between them),

• metadata describing the properties of the

learning objects.

Knowledge is represented on different levels of

abstractization. On the lower level are the LOs,

defined as entities which may be used, reused or

reffered in the learning process specified previuosly.

These are logical containers which represent

resources deliverable through the web, like lessons

(HTML pages), a simulation (Java applet), a test

(HTML pages with evaluation forms) or any other

object provided through web having learning as

goal.

Metadata is a collection of attributes of the

objects from the previous level, which are describing

the object type (text, slide, simulation, questionnaire

etc.), the requiered educational level (highschool,

university etc.), language, interactivity level etc.

The third level of abstractization (ontology) is

used for the specification of the domain concepts

and the relations between these. A domain concept

can be represented by one or more LOs (having

different attributes).

The main relationships between concepts are: Is_

part_of and Required_by dictating the hierarchical

relationships between concepts as well as the

constrains defining the mandatory learning order of

the concepts; the relation Suggested_Order can be

added optionally. The link between the concepts and

the learning objects is explicitated by the relation

Explained_by.

In order to develop the ontology for “project

management” an internationally recognized standard

was needed, to provide foundation for the definition

of the domain concept and project manager

competences. The standard was ICB - International

Competence Baseline al IPMA (Project

Management Association). The ontology of the

project management course developed by SinPers

project contains 201 concepts and the three types of

relationships mentioned above. Figure 4 presents a

fragment from the domain ontology diagram,

representing the course module “general knowledge

about a project” (Bodea, 2007).

The course ontology has been extended with the

competences ontology, taking into consideration that

a competence involves learning / proving knowledge

referring ‘n’ basic concepts. This approach is

another original element of the project. The

competency ontology (in line with ICB) allows the

identification of a possible gap between the

reference and the actual competency profiles and the

identification of the project management training

requirements. A project management learning

approach based on ontology allows finding the most

suitable training when there a similarity but does not

an exact match between training offers and the

competency gap.

5 CONCLUSIONS

The SinPers research proved that in order to

personalize services and content for e-learning

systems there are needed new solutions for at least

three major areas: design of the teaching-learning

process (actors, activities, conditions, events etc.),

creation and maintenance of an individual model of

each learner (educational requirements, preferences,

knowledge level, pre-requisites etc.), an new

structuring and access mode for the digital content

(domain and competences ontology, learning

objects, metadata).

Within a teaching-learning defined process, a

personalized unit of learning (course, lesson,

module) is composed by an activity and educational

objects tree offered to the learner. These objects are

selected from a digital content warehouse by

comparing metadata with the characteristics and

preferences of the learner and set up in a sequence

according to the relations between concepts and the

activity flow previously defined.

WEBIST 2008 - International Conference on Web Information Systems and Technologies

368

Figure 4: Project management domain ontology - fragment.

REFERENCES

Attwell, G. (2007) Personal Learning Environments - the

future of eLearning? Available from:

http://www.elearningpapers.eu/

Bodea, C., (2007) Ontology-Based Learning in Project

Management, ECEL-Electronic Journal of e-Learning,

Academic Conferences Limited, Curtis Farm,

Kidmore End: England

Dondi, C. (2007) The underground rivers of innovative e-

Learning: a preview from the HELIOS Yearly Report

2006/0.

Available from: http://www.elearningpapers.eu/

Fisher, G. (2000) Lifelong Learning - More

Than Training. Available from: http://

l3d.cs.colorado.edu/~gerhard/papers/lll99.pdf

IMS Global Learning Consortium (1999-2007). Available

from: http://www.imsglobal.org/

Martinez, M. (2000) Designing Learning Objects to

Personalize Learning, Available from:

http://www.reusability.org/read/chapters/martinez.doc

Trandafir, I., Borozan, A-M. (2007) eLearning Design

based on Personalization Requirements in

Proceedings of International Technology, Education

and Development Conference (INTED 2007), held in

Valencia, Proceedings CD

LEARNING PERSONALIZATION - Design Solutions in an e-Learning System

369