=Paper=
{{Paper
|id=None
|storemode=property
|title=Learner Model's Utilization in the E-Learning Environments
|pdfUrl=https://ceur-ws.org/Vol-924/paper16.pdf
|volume=Vol-924
|dblpUrl=https://dblp.org/rec/conf/balt/VagaleN12
}}
==Learner Model's Utilization in the E-Learning Environments==
https://ceur-ws.org/Vol-924/paper16.pdf
162
Learner Model's Utilization
in the e-Learning Environments
Vija VAGALE and Laila NIEDRITE
Faculty of Computing, University of Latvia, Raina boulv. 19, Riga, Latvia
vija.vagale@du.lv, laila.niedrite@lu.lv
Abstract. In the field of personalized systems big role is granted to the adaptive e-
learning environments. The task of these systems is very important and
complicated. With their participation the learner gains exactly the knowledge he
needs most, and the system adapts to user needs, expectations and his individual
features. In this kind of systems information about learner is saved in the learner
model also known as the user model and student model. For the system to be able
to perceive and analyze user activities correctly, is necessary to define what kind
of information about the learner has to be saved. The article gives an overview
about already existing learner models, their utilization methods and also it offers
their comparison according to various criteria.
Keywords. E-learning, adaptation, learner model
Introduction
When the tempo of life becomes more and more intensive, the necessity of the new and
effective solutions in different scopes including education arises. One of the most
actual tasks for educational quality improvement is the utilization of an e-learning
system. One part of such learning environments is passive and used only to supply
users with static content and to ensure identical system reactions to the users’ activities.
The other part of the learning environments adapts to the user as a personality by
offering learning in the most appropriate way for him. The abovementioned
environments are called adaptive learning environments (ALE). One of their tasks is to
find out user’s personal qualities that influence his learning process and his knowledge
level in certain moment of time to offer a learner certain learning content and learning
methods, which are the most appropriate exactly for him, to ensure the best learning
results. Several adaptive systems are offered in the works of Hauger and Köck [14].
The role of adaptive systems nowadays gets bigger every year. This type of
systems can help to acquire knowledge at schools, universities and other kind of
learning institutions. ALEs can be used for the student teaching in schools as a
secondary instrument to acquire fundamental and additional knowledge, but for
students in universities they serve as a primary instrument for acquiring new
knowledge and organizing their study plan. For the lifelong learning of adults ALE can
serve as an instrument for gaining new knowledge and raising professional
qualification. ALE can be used also to ensure learning interaction with already existing
learning environments. For example, ALE cooperating with popular social networks
V. Vagale, L. Niedrite / Learner Model's Utilization in the e-Learning Environments 163
can offer some specific knowledge for certain user group. Also, for the children of
preschool age adaptive learning system can be used as an instrument of gaining basic
knowledge that would be necessary at school.
The aim of the work is to explore structure models of the adaptive systems
(including learning systems), especially – user model and its components, and to
explore data which are already included in user model and select the most common
used data which would be useful for creating adaptive system based on user model.
This work is written based on scientific articles that reflect the newest trends in
formatting and utilization of the learner model. The most-cited literature has been
analyzed, as well as the newest review articles about learner model and papers that
describe already realized adaptive system examples.
Alenka Kavcic [16] points out several most relevant questions, which should be
considered when creating a user model: a) what kind of information must be included
in the user model, b) how to gain this information, c) how to represent information
about a user in the system, and d) how to create and restore a user model.
The first section of this paper gives an overview of the models used in the
adaptation process, user profiles and explanation of the learner model concept, their
common and different features, the ways to obtain profile data and its utilization for
creating a user model. In the second section an overview of the data included into the
user model of an adaptive learning system, and the data included into the learner model
are summarized. Also, a definition of this data is given. In the end of the section LM is
offered a table that summarizes employed data with the aim to obtain the most
significant data that must be included into learner model. The third section describes
the creating stages and construction techniques of the user model of an adaptive system.
The paper ends with conclusions on the accomplished overview of the learner model
utilization in adaptive systems.
1. Learner Model Essence
1.1. A Review of the Models Necessary for Adaptation
Adaptation in the learning environments is based in the well-organized models and
processes. Data that describes knowledge in the system and learner behavior are
extensive. When exploring scientific articles [8, 9, 16, 24, 28, 33] on different types of
adaptive systems, one may conclude that they are based on the three main models: a
domain model, a user model and an adaptive model.
A domain model includes two main parts: content or system offered domain
knowledge, and a supply system for this content. In [9] authors points out that a domain
model works like a data repository, which consist of topics, content, pages or nodes and
navigation links that connect the represented data design structure. Domain knowledge
consists of knowledge basic elements such as concepts, topics, knowledge items,
learning goals, learning results. Domain delivery system must support all course types
and manage to adapt to the different requirements for the course content.
An adaptive system adapts to users’ needs that is why one more important
component of these systems is a user model. In the learning systems it is also called a
student model or learner model (LM). In [26] authors call a LM the key and core of the
adaptive system. The learner model keeps all information about the learner, i. e.,
164 V. Vagale, L. Niedrite / Learner Model's Utilization in the e-Learning Environments
personal information, his knowledge, skills, and behavior in the system. Intelligent
Tutoring Systems (ITS) learner model is also called a student model [35].
Domain and learner models are connected with the help of an adaptive model. An
adaptive model ensures the application of the system flexibility theory by combining
domain and learner models [9]. By analyzing learner model student needs are gained,
and knowledge representative nodes are offered to him the system. These nodes can be
classified by knowledge type as follows: basic knowledge (includes knowledge about
definitions, formulas, etc.), procedural knowledge (solves relationship between stages),
and conceptual knowledge (refers to relationship between concepts by developing
bigger common scene) [39]. In widely spread Adaptive Hypermedia Systems (AHS) an
adaptive model is also called an interaction model [11, 27]. In the Intelligent Tutoring
Systems (ITS) adaptive model functions are fulfilled by pedagogical model [35].
Depending on the adaptive system type, in addition to the previously mentioned
models, there can also be other models, which ensure the system supplied services.
AHS systems have also the fourth model: media model [3, 28]. Along with the learner
model there is also a group model [33], which is similar to learner model but is filled
dynamically and is based on learner group identification after some common features
and behavior.
1.2. Basics of Learner Model Creation – A User Profile
To make an adaptive system, which could respond exactly how the user wants,
information about the user is needed. The easiest way how to obtain information about
the system user is to use his data from the user profile.
In the profile static (constant) information about the user without any additional
description or interpretation is kept [30]. Profile data contains learner personal data as
well as data on his individual features and habits. User data in the profile is represented
as attribute pairs – key-value. User model creation, modification and maintenance
process is called user profiling. A system should provide profile attribute initialization,
adding, saving, modification, deletion and extraction.
Unlike the profile, a user model is an abstract representation of the system user
[23], where, in addition to the profile data, some specific information about the person
is included. For example, in [27] a learner model consists of the domain independent
data that consists of Generic profile, Psychological profile, and domain-dependent data.
The user model contains all information that the system has on the user and maintains
live user accounts in the system [33]. In the general case, the profile concept is
narrower than the user model concept. In a simplified case, they can coincide
[Kules2000]. User profile data can serve as the base for the creation of the user model.
A profile keeps static information, but the user model keeps both – static and dynamic
information.
1.3. Obtaining Data for the Learner Model
When the user interacts with the system for the first time, a user profile that contains
the basic information about the user is being created in the system. Information in the
user profile can be obtained in similar ways.
V. Vagale, L. Niedrite / Learner Model's Utilization in the e-Learning Environments 165
Real people
Computer Interface
Abstract people
User profiling
Learner Model Adaptive Model Domain Model
User profile Domain Specific Adaptation Rules Course Content
Data
User modeling Domain Independent Instructional Rules Delivery System
Data
Adapting Adaptive e-Learning system
Fig. 1. Adaptive e-Learning system scheme
There are several approaches to create a user profile:
a) A user creates his profile on his own, based on his interests [25]. A part of the
user profile information can be obtained directly from the user registration form or
questionnaires: for instance, birth date and gender. For example, in [7] EASEL project
with the help of AHS Questionnaire Servlet the system asks the user some simple
questions to gain data about visual, audio, read/write and kinesthetic attributes.
However, such information as user preferences is very hard to gain and that is why they
must be taken from the user interaction with the system.
b) A system creates a profile by itself by collecting necessary information about
the user indirectly, for instance, from activity log files, where it is written what a user
has chosen and what actions he has accomplished in the system [25].
c) Mixed approach, when one part of information is input by the user, but the
second part of the information the system gains indirectly [25].
d) ALE integration with other informational systems with user data import from
some other system:
• From the informational system (IS) which ALE is related to. For example,
ALE has “cooperation” with administrative system of the educational
institution, which contains general information about an IS user. IS imports
user data into the adaptive learning environment and ALE uses this data as an
entrance data to create the first notion about the learner. When gaining data
from IS to the ALE in this way, such information as the type of knowledge
student must acquire (for example, registration to the certain course) is also
often indicated [35].
• From other type of systems with user registration. Nowadays social networks
where people spend a lot of time (for example, facebook.com, draugiem.lv)
have become widely used and popular. Versatile information about the user is
saved there, i.e., his personal data, interests, skills of communication with
people, activities in groups, etc. From that kind of systems, which widely
characterize its user, data can be taken and integrated into united adaptive
learning system user model. In this case, it would be important to anticipate
both system cooperation opportunities.
e) Gaining data from ePortfolio: a web-based electronic material resource, which
contains material collection that is made and managed by user [5, 42]. ePortfolio also
indicates the user learning or professional growth. Fsor example, in [34] a research
166 V. Vagale, L. Niedrite / Learner Model's Utilization in the e-Learning Environments
about ePortolio integration with Learning Management System (LMS) Moodle is
described.
2. Learner Model Data
2.1. Learner Model Data Types
Information included in the user model can be grouped differently: (a) by data
dependence from the subject, (b) by data obtaining type, (c) by data availability for the
learner, and (d) by data life-cycle in the user model.
Relative-to-subject information can be “domain-dependent” and “domain-
independent”. Domain-dependent information shows the knowledge level and ability
of the learner at the certain moment of time. Domain-independent information is not
dependent on the offered content; it is for example, motivation, skills, learning style.
In educational AHS (Brusilovsky [4]) LM data is divided into two big groups:
domain-specific information (DSI) and domain-independent information (DII).
Domain-specific information contains student knowledge model that describes student
knowledge level, insight about knowledge, learner mistakes and records about learning
habits and ratings. The domain-independent information includes information about the
learner skills based on his behavior. DII includes also learner learning goals, his
cognitive abilities, motivation, background, experience and preferences. The work [27]
also has similar data included in the user model division: domain-independent data
(DID) and domain-dependent data (DDD). Domain-dependent data stores specific
learner knowledge from the domain that system concludes about the learner. Domain-
independent data includes two elements: the Psychological Model and the Generic
Model of the Student Profile [22]. Psychological data are connected with the student
exploration and emotional aspect. The Generic Model of the Student Profile keeps user
interests, common knowledge and experience.
Another LM division is based on the way how data about the learner is gained:
“content-based” or “collaborative”. In content-based case, data are collected or
concluded about the learner only from his interaction with the system. In collaborative
case, data is obtained from other similar learner groups that share, for instance, similar
interests and necessities [40], and is used for a certain learner.
The user model data can be “visible” and “opaque”. Visible data can be changed
by user with the help of questions-answers offered by the system. On the contrary,
opaque LM data are not available for the user [19].
In [25], [26], [40] the division of data by its life-cycle in the system used in the
learner model or by learner interaction with the system (values are changing or not):
• Static data is data that are not changed during the student and system
interaction;
• Dynamic data is data that changes depending on the student learning progress
and interaction with the system.
Static data types are personal, personality, cognitive, pedagogical and preference
data [13]. Static data is collected either once at the beginning of system utilization or
after a determined period. This data stays unchanged during the system utilization.
Dynamic data is gained based on learner interaction with system. Dynamical data is
divided into performance data and student knowledge data. Performance data is data
gathered from the user and system interaction, and this data summarizes the
V. Vagale, L. Niedrite / Learner Model's Utilization in the e-Learning Environments 167
information about student’s achievements during the course session and is being
continuously restored. Student knowledge data is data that describes knowledge
concepts and competences. This data set gathers information about student progress in
the course of learning.
2.2. Analysis of the Data Included in the Learner Model
In this section eight works describing adaptive systems have been considered, and data
included into user model have been analyzed. The research was complicated because in
data categories used in these works were similar by meaning but with different names
and vice versa, had similar names but category content was different. By making article
research the most important data categories used in viewed scientific works were
selected: personal data, pedagogical data, preference data, personality data, cognitive
data, history data, device, context, interests of user, interests gathered by system,
results of assesement, domain expertise, acquired knowledge, performance data,
deadline extend, student knowledge currently. Below one may observe the categories
and data included in them depending on the category occurrence in the viewed works.
Personal data is a category without which no system can do. It includes user
biographical information gained, for instance, from the registration form. This category
combines personal information, demographic information and identity data. Data
included into the personal data category is similar for all authors:
• In [13] authors include in this category student name, special accessibility
needs to course materials that the student must have; affiliation; student’s
professional activities; list of degrees and qualifications; information about
student security and access credentials.
• In [10] personal data is gender; age; language; culture; name; email; password.
• In [27] personal information is name; email; password; demographic
information is age.
• In [1] personal data category includes name; surname; age; language; media
type; login.
• In [41] as demographic information: gender; age; native language; social-
cultural parameters: formal education, family income is considered as personal
data.
• In [25] authors include in personal data category personal information: name;
age; address books; demographic information: date of birth; gender;
nationality.
Next category that is mentioned in almost every work is pedagogical data that
describes how and what to learn. This category includes programs, topics, course
collections and course sequence. In the majority of examined scientific works this
category includes also learner knowledge before adaptive system utilization, for some
authors this data was described in the personal data category. Data sets included in the
pedagogical category of each article author significantly differs, for instance:
• In [13] this category consist of learning style; learning approach; course
objectives – concept list that each student must acquire during the course
session; course evaluation; course navigation control, i.e., how the course
content is navigated.
• In [27] authors include in this category academics backgrounds, for example,
technological studies contrary to economically; qualifications: certificates;
168 V. Vagale, L. Niedrite / Learner Model's Utilization in the e-Learning Environments
background knowledge: knowledge collection that is transformed into
concepts; ability to determine qualitative, quantitative and probably user
acquired concepts and knowledge; background knowledge: knowledge
collection translated into concepts; plan.
• In [1] student knowledge level in certain concepts of the onthology-based
learner model is marked with very low, low, medium, good or excellent.
• In [3] Brusilovsky calls this data a background that contains basic knowledge
with which the user started to employ the adaptive system.
• In [29] authors include skills, knowledge, abilities and plan in this category.
• In [41] the pedagogical data category includes the learning plan.
In more than a half of scientific works data that falls into the category of
preference data is observed. This category keeps student preferences relating to the
system adaption. Majority of the preferences are taken from the student, but the
remaining part is defined by system administration. Characteristic examples of this
category are the following:
• In [13] this category includes preferred presentation format; preferred
language for content display; web-design personalization; command
personalization; personal notebook; sound volume; video speed; subtitles.
• In [27] authors include in the preference data category learner defects, for
instance, bad sight; domain of application user localization.
• In [1] authors include in this category specific data that is necessary for
organizing learning process based on learner ontology model.
• In [3] Brusilovsky explains that the system cannot calculate this data,
therefore, it is gathered directly or indirectly based on the user activities.
The next category that was distinguished is personality data. This category
gathers data that describes student as a personality: learning style, concentration skills,
collective work skills, relationship creating skills, individual features and attitude
towards learning. A part of these data can be gained via tests. All in all, authors in this
category have included similar data, for example:
• In [13] authors combine in this category personality type; concentration skills,
where the base is average time that is used for learning; collaborative work
that characterizes student skills to work in groups; relational skills that
characterize student skills to communicate with teacher.
• In [27] authors include in personality data category learning style; information
reception abilities – cognitive capacities; traits of personality: introvert,
extravert; activity; inheritance of characteristics that classifies users, so that
further system could create learner stereotype models.
• In [41] this category includes specific data – emotional state, it is ability of
adaptive system to model determined user emotions with purpose to make
system behavior correction.
• In [29] authors include attitude in personality data category.
For successful adaptive system utilization it is necessary to have data on user
experience in work with computer, certain software and adaptive system. This data is
described by the half of all authors, the data type are gathered in the category system
experience, for instance:
• In [13] authors keep in the user model experience level that describes
student’s ability to work with an e-Learning system and student’s experience
in computer utilization.
V. Vagale, L. Niedrite / Learner Model's Utilization in the e-Learning Environments 169
• In [3] Brusilovsky defines experience as how familiar the user is with system
and how easily he orients in it.
• In [27] this category is called aptitude.
In the half of the reviewed scientific works the category goal or motivation is
presented. This is data about the system user long-term interests [25]; data that
characterizes reasons why the learner does some actions (for example, searches and
uses specific information) [10].
Cognitive data category describes what reference types a student has. These
characteristics can be obtained by tests. This data has an important role in the system
adaption ability to the learner. These data types are found only in three of eight
reviewed works. For example, in [13] this category includes cognitive styles. Authors
in [10] describe the data that characterizes the way of how the user processes the
information.
Below are listed categories which include data used in several works only. History
data category includes data that contains information about user activities in the system.
For example in [10] information about the last user interaction with system (log file) is
considered. In [27] authors use data about access of each page. In [1] authors include in
history data category the data on learner navigation during resource learning process.
Device category incorporates data that characterizes user environment during the
adaptive learning system utilization. In [10] it is hardware; screen size; download speed.
In [27] this is data connected with the user environment, for instance, screen resolution.
Context category incorporates data which characterizes the user access place. It is
relevant in cases when someone is using different devices to access the adaptive system.
This data intersects a little with the device category. In [10] in this category authors
keep information about the access environment, for example, access from home or
from educational institution. In [41] wider information is included in this category: user
location, time, physical and social environment, used devices, etc.
In the viewed papers Interests category is mentioned. For one part of authors those
are interests that the user indicates himself, for the other – interests collected by the
system based on the user activities. For better understanding of the observed interests,
this category was divided into two parts: Interests of user and Interests gathered by
system. In [10] the system collects interests from user keywords and searching results.
In [27] authors examine person’s interests, which are used to adapt navigation and
content.
For saving learner knowledge data categories that characterize specific knowledge
types are used. Results of assessment category contains the learner knowledge test data,
for example, [27] keeps data about all tests and exercises. Domain expertise determines
knowledge in topics that the user has interest for [10]. Knowledge acquired describes
learner knowledge in the certain moment, for example, in [27] it is mentioned that the
learner knowledge is transformed into concepts.
Summary about the examined data categories of user model is shown in Table 1.
Table column names correspond to the researched articles, and rows – to data
categories. Category utilization in certain article is represented with a “+”.
When analyzing the results of Table 1, it is obvious that personal and pedagogical
categories are the most common in the learner model. In the works of several authors
there is no precise borderline between categories personal and personality and the same
data is included into personal and personality categories. Cognitive data is similar to
personal and personality data, where data that characterizes learner and significantly
affects learning process is saved. One of the widest and most important categories of
170 V. Vagale, L. Niedrite / Learner Model's Utilization in the e-Learning Environments
the user model is pedagogical category, which incorporates learner basic knowledge
and learning plans.
Table 1. The frequency of the learner model data category utilization in scientific works
Data category type [29] [3] [25] [41] [13] [10] [27] [1]
Personal data + + + + + + +
Personality data + + + +
Cognitive data/style + + +
Pedagogical data + + + + + + +
Preference data + + + + +
History + + +
Device + + +
Context/Environment + + +
Interests of user + + +
Interests gathered by + +
system
Goal/Motivation + + + +
System Experience + + + +
Domain Expertise +
Results of assesment + + +
Knowledge acquired + +
Deadline extend +
3. Learner Model Modeling
An adaptive system continuously collects data about the learner. This process is called
user modeling and it is quite complicated. In this process the activities in LM,
mechanisms used in modeling LM and the way in which LM saves data must be taken
into account.
3.1. Formation Stages of the Learner Model
First, the adaptive system initializes the user model, and only after that data is being
refreshed in the LM. Several authors highlight one more stage that involves learner
data mining and concluding. Extensive research about the educational data mining is
found in the review article [38].
After making article analysis, the authors of this paper concluded that the most
widespread learning model formation stages are the following:
• Initialization – information and data gathering about the user, and user profile
formation that is based on obtained information. During the initialization
process the structure of the user model is defined as well as reasoning methods,
and memory (i.e., information about user abstraction state in a certain moment
of time) in the user model. There are two ways how to obtain data: explicit
questions and initial tests. This stage is used in the articles [9, 16, 25, 30].
• Updating – the system must be ensured with the learner model actualization;
the system observes user activities, evaluates user achievements, learning
progress dynamics and makes the reflexive link analysis from the user’s
V. Vagale, L. Niedrite / Learner Model's Utilization in the e-Learning Environments 171
interaction with the system. All these activities are executed by the system
implicitly or explicitly. This stage is covered in the articles [9, 16, 25, 30].
• Reasoning (i.e., extraction of the new information about the user from the
existing available data) [13]. Data mining utilization to ensure adaptation is a
new research direction which is reviewed in [21, 38].
3.2. LM Construction Techniques
The user modeling technique describes how the user model is created and maintained.
There are many techniques for modeling the user model and supplementing data in it.
For example, there are such LM construction techniques as: a stereotype model, an
overlay model, a combination model, a perturbation model, a plan model. The oldest
and the most frequently used ones are a stereotype model, an overlay model and a
combination model [6].
A Stereotype model is often used for LM to define some default values. In case of
the learning system, users are divided in the system-offered categories, i.e., stereotypes.
It has been introduced by Elaine Rich. A stereotype is a simple collection of the
aspect–values that describes the system user groups [36]. The benefit of it is that with a
small amount of information it is possible to conclude a lot of new assumptions about
the user. The stereotype review and utilization examples are given in articles [3, 6, 7,
12, 15, 20, 30, 37].
An Overlay model is widely used in adaptive hypermedia systems. A student
knowledge model is being constructed based on the concepts. A user model is restored
based on the user progress in the system. This model allows creating knowledge about
a student in each topic in a flexible way. The modeling approach of the overlay model
is based on the domain model, which is often constructed with the help of the
knowledge network or the knowledge hierarchy tree. This method requires dividing
domain model in different topics and concepts [3]. Nowadays domain model can be
built with the help of ontology [30]. The overlay model review and utilization
examples are provided in articles [6, 7, 15, 27, 30].
A Combination model employs both of the previously mentioned models. First of
all, students are divided by stereotype, and then this model is gradually modified into
an overlay model. This model is used for educational AHS [6].
A Differential model is described in [30]. It is another version of overlay model,
where the knowledge that a student must acquire in a certain period of time is
represented (i. e., expected knowledge). This knowledge can be considered as the
knowledge that is missing.
A Perturbation model. Some models are not interested in learner mistakes caused
by wrong perception or lack of knowledge. This model represents learner knowledge as
an overlay model plus his mal-knowledge [30].
A Plan model incorporates successive student actions for achieving certain goals
and desires [30].
3.3. User Data Modeling Methods
Static data elements are modeled with Attribute-Value Pairs [25]. Attributes are terms,
concepts, variables and facts that are important for both the system and the user. Their
values can be of the following types: boolean, real or string. For modeling uncertain
information elements like user knowledge more difficult approaches there are used
172 V. Vagale, L. Niedrite / Learner Model's Utilization in the e-Learning Environments
such as rules with certainty factors, fuzzy logic, Bayer probability networks or
Dempster-Shafer theory of evidence [16]. User condition-based language modeling
approach is applied to the dynamical data elements, where a connection between the
provided service and a context is based on if-then logic.
To represent the relationship between data elements the hierarchy tree modeling
approach and ontology are used. The user ontology is described in following articles:
[1, 13, 29]. In [43] a research about e-learning platform knowledge management with
the help of ontology is described. Systems used for user modeling are:
• UMT [2] that allows developers to define hierarchically arranged user
stereotypes, rules of the user model conclusions and contradictions.
• PROTUM [44] represents the user model content as a list of constants, where
each constant is attached to a certain type and confidence factor. This tool has
deeper stereotype retraction mechanisms than UMT.
• TAGUS [32] represents assumptions about the user with first-order formulas
by indicating different types of assumptions. This tool allows defining
stereotype hierarchies and contains conclusion mechanisms.
• UM [17, 18, 19] toolkit3 includes user modeling by indicating user knowledge,
views, desires and other user characteristics with attribute-value pairs.
• BGM-MS [19] user or user group stereotype choice is based on the
assumptions that are gained using predicate logic. User knowledge defining is
based on conclusions that are obtained employing different assumptions.
• DOPPELGÄGER [31] is a server that collects information about the user with
hardware and sensor software. A user can visualize, check and edit his data.
4. Conclusions
Adaptive system range is wide and there are a lot of researches in this scope including
the user model utilization.
During the analysis of the adaptive system structure authors concluded the
following: (1) depending on the type of the adaptive system, model names included in
it may differ but their essence and tasks remain similar; (2) each adaptive learning
system must have at least three components: (a) a domain model for keeping system-
offered knowledge; (b) a learner model (user model, student model) which describes in
an understandable way for the system a person sitting in front of the computer and
willing to learn; (c) an adaptive model (interaction model) with the help of which
system-offered knowledge is delivered to the learner in an understandable way.
One of the most important adaptive learning system components is the learner
model. It includes data from the user profile, however, when making a good adaptive
system learner model can also include additional data characterizes the learner in a
more comprehensive way.
It would be recommended to divide all data included in the learner model into
some basic categories:
• Personal data, where data about the personality identity is stored (name,
surname, login, password, language, gender, date of birth).
• Personality data – data that characterizes the learner as personality (individual
features, learning style, concentration skills, personality type, collective work
skills, emotional situation, attitudes).
V. Vagale, L. Niedrite / Learner Model's Utilization in the e-Learning Environments 173
• Pedagogical data – data which characterizes anything that a learner must learn
(programs, themes, course sequence, plan).
• Preference data – data that adapts working environment for learning (language,
presentation format, sound value, video speed, web design personalization).
• System experience – data that characterizes learner’s earlier gained experience
with computers and software used in the learning process (obtained
certificates, skills in e-Learning system utilization).
• Cognitive data – data that represents reference types of the learner.
• History data – data about all learner’s activities.
• Device data – data that characterizes working environment of the system user
(hardware, download speed, screen resolution); learner’s location, time; and
devices used.
• Student knowledge at the current moment of time – data that describes student
knowledge gained in the learning process.
Future work would be some kind of practical realization of the learner model using
obtained results about data included in the learner model and their types.
References
[1] Behaz, A. and Djoudi, M. Adaption of learning resources based on the MBTI theory psychological
types. IJCSI International Journal of Computer Science Issues, 9, Issue 1, No 2, (2012).
[2] Brjanik, G. and Tasso, C. A Shell for Developing Non-monotic User Modeling Systems. International
Journal of Human-Computer Studies 40 (1994), 31-62.
[3] Brusilovsky, P. Methods and techniques of adaptive hypermedia. Spec. Issue On Adaptive Hypertext
and Hypermedia, User Modeling and User Adapted Interaction, 6, 2-3 (1996), 87-129.
[4] Brusilovsky, P. The construction and application of student models in intelligent tutoring systems.
Journal of Computer and System Scences International, 32, 10 (1994), 70-89.
[5] Challis, M. Portfolio-based learning and assessment in medical euducation. AMEE Medical Education
Guide No.11. 1999.
[6] Colan, O., Dagger, D., and Wade, V. Towards a Standards-based Approach to e-Learning
Personalization using Reusable Learning Objects. In World Conference on E-Learning in Corp., Govt.,
Health & Higher Ed. AACE, ( 2002), 210-217.
[7] Colan, O., Wade, V., Bruen, C., and Gargan, M. Multi-Model, Metadata Driven Approach to Adaptive
Hypermedia Services for Personalized eLearning. In Second International Conference on Adaptive
Hypermedia and Adaptive Web-Based Systems ( 2002), 100-111.
[8] De Bra, P. and Ruiter, J. P. AHA! Adaptive hypermedia for all. In WebNet'2011, World Conference of
the WWW and Internet, AACE, ( 2001) 262-268.
[9] Esichaikul, V., Lamnoi, S., and Bechter, C. Student Modelling in Adaptive E-Learning Systems.
Knowledge Management & E-Learning: An International Journal, 3 (2011), 342-355.
[10] Frias-Martinez, E., Magoulas, G., Chen, S., and Macredie, R. Automated User Modeling for
Personalized Digital Libraries. International Journal of Information Management, 26, 3 (2006), 234-
248.
[11] Garcia-Barrios, V. Personalisation in Adaptive E-Learning Systems. A Service-Oriented Solution
Approach for Multi-Purpose User Modelling Systems. Dissertation in fulfilment of the requirements for
academic degree, http://www.iicm.tu-graz.ac.at/iicm_thesis/vgarcia.pdf, (2007).
[12] Garlatti, S., Iksal, S., and Kervella, P. Adaptive On-Line Information System by means of a Task
Model and Spatial Views. In 2nd Workshop on Adaptive Systems and User Modeling on the WWW
(1999), 55-69.
[13] Gomes, P., Antunes, B., Rodrigues, L., Santos, A., Barbeira, J., and Carvalho, R. Using Ontologies for
eLearning Personalization. In 3rd Learning Conference, Portugal (2006).
[14] Hauger, D. and Köck, M. State of the Art of Adaptivity in E-Learning Platforms. In 15th Workshop on
Adaptivity and User Modeling in Interactive Systems Halle/Saale, Germany (2007), 355-360.
[15] Hewagamage, K. P. and Lekamarachchi, R. S. Learning Patterns: Towards the Personalization of
ELearning. In 5th International Information Technology Conference ( 2003).
174 V. Vagale, L. Niedrite / Learner Model's Utilization in the e-Learning Environments
[16] Kavcic, A. Fuzzy user modeling for adaptation in educational hypermedia. IEEE Transactions on
Systems, Man, And Cybernetics, 34, 4 (2004).
[17] Kay, J. UM: A Toolkit for User Modelling. In Second International Workshop on User Modeling
(1990), 1-11.
[18] Kay, J. A Scrutable User Modelling Shell for User-Adapted Interaction. Sydney, 1999.
[19] Kay, J. The um Toolkit for Reusable, Long Term User Models. User Modeling and User Adapted
Interacton: The Journal of Personalization Reseach 4 (1995), 149-196.
[20] Kay, J. User modeling for adaptation. In Stephanidis, C., ed., User Interfaces for All: Concepts,
Methods, and Tools. Florence, 2000, 271-294.
[21] Khirbi, M., Jemni, M., and Nasraoui, O. Automatic Recommendations for E-Learning Personalization
Based on Web Usage Mining Techniques and Information Retrieval. In Educational Technology &
Society (2009), 30-42.
[22] Kobsa, A. Generic User Modeling Systems. In User modeling and user-adapted interaction. (2001),
49-63.
[23] Koch, N. Software Engineering for Adaptive Hypermedia Systems. Verlag Uni-Druck, Munich, (2001).
[24] Kules, B. User Modeling for Adaptive and Adaptable Software Systems. In ACM Conference on
Universal Usability,Arlington (2000).
[25] Liu, H., Salem, B., and Rauterberg, M. A survey on user profile modeling for personalized service
delivery systems. In IADIS International Conference on Interfaces and Human Computer Interaction
( 2009), 45-51.
[26] Li, Q., Zhong, S., Wang, P., Guo, X., and Quan, X. Learner Model in Adaptive Learning System.
Journal of Information & Computational Science, 7, 5 (2010), 1137-1145.
[27] Martins, A., Faria, L., Vaz de Carvalho, C., and Carrapatoso, E. User Modeling in Adaptive
Hypermedia Educational Systems. In Educational Technology & Society, 11, 1 (2008), 194-207.
[28] Mulwa, C., Lawless, S., Sharp, M., Arnedillo-Sanchez, I., and Wade, V. Adaptive educational
hypermedia systems in technology enhanced learning: a literature review. In 2010 ACM Conference in
Information Technology Education, Midland, USA (2010).
[29] Nebel, I., Smith, B., and Paschke, R. A user profiling component with the aid of user ontologies. In
Workshop Learning - Teaching - Knowledge - Adaptivity (LLWA 03), Karlsruhe, (2003).
[30] Nguyen, L. and Do, P. Learner Model in Adaptive Learning. In World Academy of Science,
Engineering and Technology ( 2008), 396-401.
[31] Orwant, J. Heterogenous learning in the Doppelganger user modeling system, (1995), 107-130.
[32] Paiva, A. and Self, J. TAGUS: A User and Learner Modeling System. In 4th International Conference
on User Modeling, Hyannis (1994), 43-49.
[33] Paramythis, A. and Loild-Reisinger, S. Adaptive Learning Enviroments and eLearning Standarts.
Electronic Journal on e-Learning, 2, 1 (2004), 181-194.
[34] Queiros, R., Olivera, L., Leal, J., and Moreira, F. Integration of ePortfolios in Learning Management
Systems. In International Conference Computational Science and Its Applications - ICCSA 2011,
Santander, (2011), 500-510.
[35] Riad, A., El-Minir, H., and El-Ghareeb, H. Review of e-Learning Systems Convergance from
Traditional Systems to Services based Adaptive and Intelligent Systems. In JCIT ( 2009), 108-131.
[36] Rich, E. Stereotypes and User Modeling. User Models in Dialog Systems (1989), 35-51.
[37] Rich, E. User Modeling via Stereotypes. Cognitive Science, 3, 4 (1979), 329-354.
[38] Romero, C. and Ventura, S. Educational data mining: A review of the state-ofthe-art. IEEE
Transactions on Systems, Man, And Cybernetics, part C: Applications and Reviews, 40, 6 (2010), 610-
618.
[39] Shute, V. and Towle, B. Adaptive e-learning. Educational Psyhologist, 38, 2 (2003), 105-114.
[40] Somyürek, S. Student Modeling: Recognizing the Individual Needsof Users in e-Learnig Enviroments.
International Journal of Human Sciences, 6, 2 (2009), 429-450.
[41] Sosnovsky, S. and Dicheva, D. Ontological technologies for user modeling. Iternational Journal
Metadata, Semantics and Ontologies, 5, 1 (2010), 32-71.
[42] Van Wasel, M. and Prop, A. The infulance of portfolio media on student perceptions and learning
outcomes. In Student Moblity and ICT: Can E-LEARNING overcome berriers of Life-Long learning?,
Maastricht, 2008.
[43] Vasilyeva, E., Pechenizkiy, M., and Puuronen, S. Knowledge management challenges in web-based
adaptive e-learning systems. In 5th International Conference on Knowledge Management, Springer,
(2005), 112-119.
[44] Vergana, H. PROTUM: A Prolog Based Tool for User Modeling. WG Knowledge-Based Information
Systems, Department of Information Science (1994).