Intelligent Automation & Soft Computing

ISSN: 1079-8587 (Print) 2326-005X (Online) Journal homepage: http://www.tandfonline.com/loi/tasj20

Proposing a novel adaptive learning management

system: an application of behavior mining &
intelligent agents
Rozita Jamili Oskouei & Nasroallah Moradi Kor
To cite this article: Rozita Jamili Oskouei & Nasroallah Moradi Kor (2016): Proposing a novel
adaptive learning management system: an application of behavior mining & intelligent agents,
Intelligent Automation & Soft Computing, DOI: 10.1080/10798587.2016.1186429
To link to this article: http://dx.doi.org/10.1080/10798587.2016.1186429

Published online: 05 Jul 2016.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tasj20
Download by: [5.221.177.90]

Date: 06 July 2016, At: 10:52

Intelligent Automation & Soft Computing, 2016

http://dx.doi.org/10.1080/10798587.2016.1186429

Proposing a novel adaptive learning management system: an application of

behavior mining & intelligent agents
Rozita Jamili Oskoueia

and Nasroallah Moradi Korb

Department of Computer Science and Information Technology, Mahdishahr Branch, Islamic Azad University, Mahdishahr, Iran; bYoung Researchers
and Elite Club, Kerman Branch, Islamic Azad University, Kerman, Iran

Downloaded by [5.221.177.90] at 10:52 06 July 2016

ABSTRACT

Nowadays, the internet has become one of the main resources for distance education, learning or
training purposes. Various learning management systems (LMSs) have been designed for providing
various services to learners in all parts of the world. Each of these LMSs has some similarities or
differences with others regarding to capabilities or services. One of the main problems in Iran is teaching
foreign languages to people with different genders, experiences, jobs, educational levels, etc. People
need to learn English for different purposes, but unfortunatly, all the learning matrials are similar for all
the learners without considering their capabilities, age, actual usage, etc. This paper attempts to design
a personalized LMS for teaching foreign language to people whose native language is not English. Our
experimental results indicated that the proposed LMS is useful for teaching conversation or grammar
of foreign languages such as English language even to illiterate people or those with low education.
Furthermore, the proposed LMS decreased the time of learning process and increased the duration of
remembering the words or grammar.

1.Introduction
Learning Management Systems (LMSs) are developed to
achieve these goals. Currently, several LMSs are available and
each benefits from special capabilities for delivering knowledge or information to students or learners all-around of the
world. Therefore, an Intelligent agent (AI) is a combination of
software and hardware with some intelligent features. Each IA
has sensors for getting inputs from environments attempting
to achieve its defined goals (Oskouei, Varzeghani, & Samadyar,
2014). Currently, the AIs are used in various applications in
LMSs (Ghanilou & Oskouei, 2014).
In this study, we used behavior mining techniques to
extract each individual learners pattern with tracking her/
his interaction with LMS and AI for adding intelligent features to the available LMSs. The contents visited by learners,
selected sources or training courses, submitted assignments
and the learners personal information such as gender, age,
family background and etc. were the inputs for exploring the
patterns in the study. Finally, we provided a possibility to predict new learners favorite contents or select suitable media
for delivering course contents to the learners by forecasting
their personal preferences. Evaluation of our proposed LMS
has shown that its suitability for teaching some courses such
as foreign languages to the students or learners with special
problems in remembering pronunciation and the meaning of
new words, the old and the illiterate.
This paper is organized in seven sections. Section 2 introduces the LMSs and explains their capabilities and weakness.
Section 3 represents information on LMSs with some intelligence features. Section 4 includes the steps of creating our
proposed adaptive architecture for LMSs. Section 5 discusses

CONTACT Rozita Jamili Oskouei

2016 TSI Press

[email protected]

KEYWORDS

Learning Management
Systems (LMS); behavior
mining; pattern discovery;
personalized features;
foreign language learning

more about the proposed system. Section 6 shows experimental

results of testing the proposed system. Finally, Section 7 provides conclusions on our paper by summarizing the results.

2. LMSs, their capabilities and weaknesses

Learning Management Systems (LMSs) are also called Virtual
Learning Environments or eLearning systems. These systems
are currently used by most of the governmental or private
organizations and universities all around the world to teach
different courses and subjects to the learners (Ghanilou &
Oskouei, 2014; Ellis, 2009).
2.1. Comparing the capabilities of available LMSs
Currently, several LMSs are available with many common capabilities and some special facilities. Some of the common facilities provided by different LMSs are [http://elearningindustry.
com/the-20-best-learning-management-systems]: Course registration, catalog representation, recording the list of learners,
tracking the learners activities, discussion forums, file sharing,
managing the assignments and syllables, building reports, etc.
Generally, the learning management systems are divided
into two main categories:
Open source LMSs: These types of LMSs are benefiting from some advantages such as; free accessibility for
everyone, possibility of developing these systems based
on requirements such as adaption with special languages, etc. Some of the popular open source LMSs and
their capabilities and limitations are discussed in this
part.

Downloaded by [5.221.177.90] at 10:52 06 July 2016

R. J. Oskouei and N. Moradi Kor

Moodle: It is mainly used in schools and small institutes or colleges. It is a powerful web-based software
capable of managing teaching process of a course during a semester. Some universities use this software for
simultaneous teaching to approximately 200,000 online
students, [http://docs.moodle.org]. Some facilities of
this LMS are as follows: Providing assignments, discussion forums, downloading files, grading, instant messaging, online calendar, online news and notifications,
video conference, etc. Moreover, Moodle supports 70
languages. Furthermore, this LMS has several modules
developed by software developers for different usages.
Some of these modules are: Module dialog, Module
selection, Module exam, Module Lessons and Module
workshops.
Disadvantages of Moodle: This LMS does not have an
internal email system. Therefore, the learners must have
an external e-mail ID for creating communications.

Atutor: It is a web-based learning management system,
which is developed for the first time by Technology
Research Center, University of Toronto. It provides
facilities for learners such as; ability for managing the
contents of the selected courses, sending messages to
other members, making working groups and Weblogs,
sharing the lessons, ability to work with social networks, providing e-mail services, having graphical
editor, capability of identifying learners with highest
performance, etc. Similar to Moodle, this LMS also has
different modules.
Claroline: It is an eLearning platform created for the
first time by Belgium Louvain University. It supports 35
languages currently. This LMS has several facilities such
as: Ability to write descriptions for the course, establishing documents and files with various formats, managing
public and private forums, creating an eLearning path,
organizing working groups, creating assignments, notifying the tasks or activities, publishing announcements,
checking attendance or completed assignments, etc.
[http://Claroline.net].
Dokeos: It is developed for the first time by a company
named Dokeos in Belgium. This LMS uses Microsoft
PowerPoint and Excel software for creating contents
of each e-learning course. This LMS lacks any special
feature. Some of these facilities are: Ability to create
dialogue, having glossary, create groups, etc. [http://
Dokeos.com].
eFront: It is a LMS with easy-to-use eLearning system,
which has scientific and commercial usages. It is widely
used in the Greek ministries. It has a management system based on Ajax. Some of its facilities are: Learners
management, file management, examination, presenting assignments, course management, communication
tools, tracking the progress of learning, learners identification, etc. [http://efrontlearning.net].
Ilias: It is developed for the first time by Economics
and Business department at the University of Cologne,
Germany. In addition to common capabilities, this LMS
supports Web 2.0 features such as: RSS, podcasting
systems and Google Maps. Furthermore, it supports a
variety of Web services. Also, the SOAP structure of this
software has been extended. [http://Ilias.de].

Disadvantages of ILIAS: Un-acceptance of Windows

operating system as the server is the main disadvantage
of this LMS.
Olat: It is developed for the first time in 1999 in the
University of Zurich in Switzerland. This software is based
on Java (most of the LMSs are developed with PHP). This
system has a capability to manage up to 700 learners at a
time with the help of Linux server [http://Olat.org].
Sakai: This LMS is used by 150 institutions formally.
This software is also based on Java. Some of facilities
provided by this LMS are: Presenting announcements,
internal e-mail, and former chat tools, creating individual profile for each learner, making schedule for delivering course materials etc.
Other Open source LMSs: Other open-source LMSs are:
LAMS, Fle3 and Docebo. Docebo use cloud computing
systems.
Commercial LMSs: These types of LMSs are mainly
designed for commercial purposes. Some of these LMSs
are: Sumtotal, Saba, Blackboard, Giunti labs and Plateau.
The main disadvantage of the software is the high cost.
Furthermore, the software is not open source and free
for all the people.
Results of a study in Carleton University show that Moodle
is the most common and popular LMS, which is used by students and instructors in American universities. In Iran, four
LMSs including Moodle SakaiAtutor and Claroline have the
highest usage rates by universities, companies and the learners
[http://farinsoft.com].
2.2. Weaknesses of all available LMSs
The weaknesses of all the discussed LMSs are as follows:
These systems do not have any personalized services for
their learners.
The required time duration for learning one course is
defined in each period.
Behavior and preferences of learners are different. Based
on religion, age, social community, culture, etc., they
require some special knowledge. No one of the available
LMSs has been considered on these points.

Both LMSs and the traditional learning methods
are failing in delivering some of the courses such as
Mathematics and foreign languages to the learners based
on learners special characteristics.
Since English is currently one the biggest requirements for
most Iranians in all ages, with any job or educational levels
for various objectives, therefore, providing a way for delivering individual contents or selecting perfect materials or media
(such as game, video, music and etc.) based on the learners
behaviors would be a big issue. We attempted to use behavior
mining technique for extracting learners individual behaviors
or preferences with the help of IAs capabilities.

3. Intelligent LMSs
There are eLearning systems with some degree of intelligence
or adaption. In this section, we attempted to discuss those
systems.

Intelligent Automation & Soft Computing

Table 1.All Categories of Contents in This Study.

No.
1
4
7
10
13
16
19
22
25
28
31

Categories
Furniture
Cinema
Economics
Sport
History
University Course
Genetics
Environment
Telecommunication
Transportation
New Models of Cloths

36
39

Animals
Shipping

No.
2
5
8
11
14
17
20
23
26
29
32
34
37
40

Categories
Cosmetics
Home Appliance
Child Care
News
Geographic
Medical Info.
Management
Business
Culture
Astrology
Agriculture
Decoration
Hotel
Military equipment and war

Downloaded by [5.221.177.90] at 10:52 06 July 2016

3.1. Adaptive Educational Hypermedia Systems (AEHS)

These systems attempt to use learners profile and add some
services based on the information. The main goals of these
systems are: Maximizing the learners satisfaction, increasing
learning speed and minimizing the learning time (Brusilovsky
& Peylo 2003, E. Popescu, 2008). Therefore, an adaptive educational system must be able to provide individual learning
paths for each learner, track each learners activities and extract
each learners learning models (including preferences, interests, domain knowledge and etc.) to facilitate each individual
learners learning process (Boticario, Santos, & van Rosmalen,
2005).Two important features of these LMSs are:
They attempt to extract the learners characteristics such
as knowledge, age, experience, background, etc. for creating a system with some degree of adaption. This step
is known as Learner Modeling or Adaptive Learner
Navigation.
Two types of adoptions are available:
Adaption in presenting the delivery of contents
Adaption in presenting the links to various subjects or
courses

No.
3
6
9
12
15
18
21
24
27
30
33
35
38

Categories
Kitchen Appliance
Multimedia/ Entertainment
ICT
Internet
Education
Health Issues
Job and Replacement
Electronics
Astronavigation
Vegetables and Fruits
Tourism
Flowers
Art


AEHS focused on students domain specific and
domain- independent information. Further, they have
several options for delivering contents to students.

ALMSs concerned on students knowledge, teaching
models and learners learning paths.
ALMSs are developed based on creating contents for
each course, modeling learners by tracing their learning
paths and finally decide the best model for delivering
contents.
No one of the proposed LMSs have facilities for teaching
special courses for people with low educational levels or illiterates, but our proposed LMS for a first time attempts to create
facilities for teaching English Conversations or grammar to
these people.

4. Our proposed adaptive LMS

Our proposed adaptive LMS is based on behavior mining application for detecting learners individual patterns along with
intelligent multi-agents application for adding intelligence and
making automatic decisions on LMSs for delivering personalized contents to the learners.

3.2. Intelligent Tutoring Systems (ITS)

4.1. Creating pattern for each individual learner

Intelligent Tutoring Systems (ITSs) are software designed for

learning purposes and have an Artificial Intelligence (AI) component (Ghanilou & Oskouei, 2014). These systems, for each
individual student, will trace the learning path and activities
to detect each individual students knowledge level and design
the suitable adapted instruction and provide facility for the
students to send their feedbacks. Moreover, each Intelligent
Tutoring System (ITS) has four modules: Knowledge Base
Model, Learners Model, Teaching Model and Control Engine.

Data mining is the process of extracting knowledge and interesting patterns from a huge amount of data. Pattern recognition
searches to discover interesting association rules or correlations
between various attributes of entities (such as users) in relational or transactional databases (Jiawei, Kamber & Pei, 2011).
In this research, in the first step, 200 English language learners activities and transactions in one LMS (Moodle is used in
our research work), including their course activities and visited
references, exercises and passed examinations, their performance were collected for three continues semesters. After the
competition of data collection, the following steps were done
for creating each individual students learning pattern.

3.3. Adaptive Learning Management Systems (ALMS)

Adaptive Learning Management Systems (ALMSs) are using
learners preferences and learning styles for delivering personalized contents of the courses (Radwan, 2014). ALMSs have
three core elements (Oxman & Wong, 2014): Content model,
A Learner Model, An Instructional Model.
3.4. Similarities and differences between our proposed
LMS with available LMSs (AEHS, ITSs and ALMSs)
Most of the available developed LMSs are concerned on:

Pre-processing steps: All records with unsuitable, noisy

or missing data were removed.
In the second step, one item set for each session of transaction done by each individual student/learner was created. These item sets contain the activities of students
in LMS, including the visited subjects, uploaded assignments, feedbacks, etc.
In the third step, one DB for each learner was made,
and then, the frequent item sets for each student/learner
with applying Eclat algorithm were extracted.

R. J. Oskouei and N. Moradi Kor

Downloaded by [5.221.177.90] at 10:52 06 July 2016

Figure 1.Architecture of the Proposed Intelligent LMS.

In the fourth step, one integrated behavior pattern for

each student/learner in LMS was created as a virtual
Learning environment. For this purpose, all frequent
item sets extracted for each individual student/learner,
and one sequential pattern was obtained. In this pattern,
all items were ordered by their confidence.
In the fifth step, the main goal was to group the learners
based on their behavior patterns and extracting association rules (we used FP-Tree (Han & Pei, 2000) for
association rule mining) between their different characteristics or attributes such as; discovering the relationships between learners gender, age, job, education level
and their selected learning materials.
For detecting similarities, first, we assigned an integer
number for each activity done by the learner. In our
study, we assigned an integer number for each topic on
the English conversation course. For this issue, we first
defined one separate category for each topic, such as:
All lessons related to the kitchen such as cooking or preparing foods or sweets, dishes in kitchen are grouped
under the category of Cook and assigns integer number
3 for this category.
All lessons related to furniture and home decoration are
grouped under category of Home-Furniture and assigns
integer number 1 for this category.
All lessons related to farmers, fruits, vegetables and
garden are grouped under category of Agriculture and
assign integer number 30 for this category as it is presented in Table 1.
After assigning an integer number for each category,
all items in sequential pattern are converted to integer
numbers and all repeated items are removed.

With applying one of the similarity measures, such
as Euclidean Distance, Mahalanobis Distance and
Minkowski Distance, we explore the similar patterns of
learners and create clusters of learners based on the similarities of their behaviors.
In this step, the relationships or possibilities of existence
any association rules between students/learners classes

(groups) and their special attributes such as gender, age,

job, religious and etc., are exploring.
Finally, the discovered association rules and groups are
recorded. This information will be used for identifying/
predicting the new learners interest areas or topics and
propose more suitable topics for achieving better performance in learning.
Based on the results of the above studies, we made a decision
tree for predicting or commenting the suitable contents to new
learners based on measuring the similarity of their characteristics (especially gender, age and job) with the modeled learners.
4.2. Proposed architecture for intelligent LMS
For designing the intelligent LMS with the capability of applying personalization in presenting the contents for learners, we
used Prometheus methodology for creating intelligence capabilities. The Architecture of the proposed intelligent learning
management system with its intelligent multi-agents is shown
in Figure 1. The implementation of the agents in our proposed
system has done by JADE (Scutelnicu et al., 2014).
It is evident from Figure 1, that the architecture of our proposed intelligent LMS has 4 layers. More information about
these four layers is described below.
Users Layer: The main users of this system are; learners
and teachers.
Intelligent Agents Layer: Intelligent multi-agents are
used in this proposed intelligent LMS. Each one of the
agents is responsible for handling specific functions,
operations or objectives.
Overall intelligent agents that are used in our proposed
intelligent LMS are:
Interaction/ Communication Agent: It is responsible for
managing all learners and teachers communications or
transactions with LMS.
Agent for Tracking Learners All Activities: It is
responsible for tracking and recording all activities of

Downloaded by [5.221.177.90] at 10:52 06 July 2016

Intelligent Automation & Soft Computing

users such as the visited contents, uploaded assignments

and etc.
Users Pattern Extractor Agent: This agent with the
help of data mining techniques such as pattern extracting technique attempts to extract behaviors of learners
and classify users based on their behaviors. Further, this
agent has the capability for applying Association Rules
and identifying of related characteristics of learners.
Personalization Agent: It is responsible for applying
results of the pattern extractor agents for creating some
specific services for each individual learner.
Agent for Creating Personalized Recommendations:
This recommendation would be based on learners characteristics or behaviors.
Planning Agent: This agent attempts to use different
characteristics of the learners, their behaviors and their
before performance to make a new plan for improving
their performance.
Agent for Managing the Location of Contents: This
agent attempts to allocate contents with related concepts
with each other.
Applications Layer: In this part, all operational functions of the system are defined. Some of these functions
include:
Creating and implementing the training plan

Figure 2.Overall Interactions Diagram of the Proposed System.

Modeling the educational content

Evaluating the learning performance
Classification of contents
Creation/ updating the learners profile
Modeling the learners behaviors
Creating personalized recommendations
Database Layer: The proposed system databases are:
Learners, Instructors, Patterns, Contents, Learners classification, Learners feedback and Materials Information
DB.

5.Discussion
To create various categories of contents associated with exploring behaviors, selections, preferences and learning paths of
learners, we made a Taxonomic classification (hierarchical classification schema (tree)) for visited contents. In this schema,
the most visited contents are presented in the first level of tree,
and the following levels show other selections of learners. The
root of this tree can be one of our selected factors to explore and
distinguish the behaviors of learners such as age, educational
level, job, gender, etc. In the schema, each node represents the
content, while each edge indicates a weight equal to the percentage of visits by selected learners.

R. J. Oskouei and N. Moradi Kor

To establish communications between the agents, we needed

to provide facilities for exchanging data between these agents
(Huhns and Singh, 1998). To this end, we used Prometheus
methodology, which in turn uses a interaction diagram for creating communication. To plot each a transaction diagram, we
needed to have the cases of application, since each transaction
diagram defines just one feature or capability of the system.
Overall interaction diagram of the proposed system is shown
in Figure 2.

Downloaded by [5.221.177.90] at 10:52 06 July 2016

6. Experimental results
To develop behavior patterns and association rules between
various characteristics (such as gender, age, job and etc.) of the
learners, 400 learners of English conversation learning classes
were selected. These learners were in different age ranges, from
5 to 60, with different jobs and living locations. All of them
were in the first stage of learning, and no one had a history
of pervious training. Their native languages were Persian and
Turkish.
These learners started their learning with Moodle, and we
tracked all of their activities and made their frequent activities item set. We provided an optional selection possibility
for the learners so that they could select their favorite topic
for learning or follow the general topics. Interestingly, 95%
of the learners preferred to select their favorite topics to start
or continue their learning process. The initial results showed
that the categories topics selected by female and male learners
were different.
In the next step, we focused on age and job factors of the
learners. According to our results from data mining of learners
behaviors tracking, the learners selection patterns related to
their gender, age, job, education level, life style, etc. Therefore,
we made various patterns for learners and extracted several
association rules between their selections and characteristics.
Finally, we used these results for creating our intelligent LMS.
To evaluate our results and the proposed intelligent LMS,
in the first step, these learners were divided into two groups.
One group was trained with simple LMS (such as Moodle)
and the other group started their training with our proposed
intelligent LMS. Our system, at the time of registration asks the
learners to fill a form and must enter the required information
regarding their gender, age, job, etc. at the time of registration.
In the next step, the system will search for the suitable matching
pattern for every new learner based on their characteristics.
Then, the system would suggest them special topics for start
and a learning path based on the similarity of their characteristics with our ex- learners.
The results indicated that the learners under training with
our proposed system had more satisfaction and higher learning
performance, especially based on their feedback and ability for
remembering words. We tested the proposed intelligent LMS
only on English Conversation learners. Since, learning a foreign language, especially remembering the new words is a very
difficult task for all students, and as in Iran, the students do not
benefit from daily speaking of their foreign language learning,
thus one of the main problems for such students would be
remembering words and using them properly. As our system
tries to propose topics for learning based on the students preferences, lifestyle or some daily basic requirements, the students
will be able to recall these words frequently in their mind during the day, and finally get better results.

7.Conclusion
In this study, behavior mining techniques were used to extract
the students patterns, preferences, learning styles and finally
to discover a method for delivering suitable and personalized
contents to students based on their special characteristics.
Association rule mining techniques (FP-Tree) were used for
exploring the relationships between various factors or characteristics. We also used intelligent multi-agents for creating
intelligence features.
Our experimental results showed that proposed LMS have
capabilities for extracting learners behaviors with applying
behavior mining techniques. Further, the proposed LMS delivers personalized contents based on the extracted learners patterns of learners. Examining the proposed LMS is revealed that,
it decreased the time of learning process and increased the
duration of remembering the words and grammar. Moreover,
our results indicated the improvements in performance and
improved the percentage of students satisfaction. Adults, children and people with low educational levels were the target
learners in this proposed LMS and it obtained their maximum
percentage of satisfactions.

Disclosure statement
No potential conflict of interest was reported by the authors.

Notes on contributors
Rozita Jamili Oskouei received a Ph.D. from the
Department of Computer Science and Engineering,
Motilal Nehru National Institute of Technology
(MNNIT), Allahabad, India. Currently she is assistant professor at the Mahdishahr Branch, Islamic
Azad University, Semnan, Iran. She is Chief Editor of
International Journal of Computer and Information
Technologies (IJOCIT). Her research areas are
mainly focused on; behavior mining, Web mining
and various applications of Data mining in medical, education and intelligent agents. She is interested in developing learning management systems through intelligent agents. Moreover, she is currently studying
Intelligent Transportation Systems (IT) and developing Smart cities.
Nasroallah Moradi kor is Dean of Editorial Director
at the Center of Advanced Scientific Research and
Publications (CASRP) in United Kingdom. Currently
he is Ph.D. Scholar at the Research Center of
Physiology in Semnan University of Medical Science.
He is serving as editorial member of several reputed
journals. His research program is primarily focused
on the effects of early life stress influences adult
learning and memory.

ORCID
Rozita Jamili Oskouei http://orcid.org/0000-0001-7162-9373
Nasroallah Moradi Kor http://orcid.org/0000-0001-9905-6845

References
Boticario, J.G., Santos, O.C., & van Rosmalen, P. (2005). Issues in
developing standard-based adaptive learning management systems.
EADTU Working Conference: Towards Lisbon 2010, Collaboration for
Innovative Content in Lifelong Open and Flexible Learning.
Brusilovsky, P., & Peylo, C. (2003). Adaptive and intelligent web based
educational systems. International Journal of Artificial Intelligence in
Educational Systems, 13, 156169.

Intelligent Automation & Soft Computing

Downloaded by [5.221.177.90] at 10:52 06 July 2016

Ellis, R.K. (2009). A field guide to learning management systems.

American Society for Training & Development (ASTD). ASTD Inc.
Ghanilou, Maryam, & Oskouei, Rozita Jamili (2014). Enhancing
performance of Learning Management Systems (LMSs) through
intelligent agents. MAGNT Research Report, 2, 192198.
Han, & Pei. (2000). Mining frequent patterns by pattern-growth:
Methodology and implications. ACM SIGKDD Explorations Newsletter,
2, 1420.
Jiawei, Han, Kamber, M., & Pei, J. (2011). Data mining, 3rd ed. San
Francisco, CA, USA: Morgan Kaufmann Publishers Inc. ISBN
0123814790 9780123814791.
Huhns, M.N., & Singh, M.P. (1998). Managing heterogeneous transaction
Workows with co- operating agents. In N. Jennings & M. Wooldridge
(Eds). Agent technology: Foundations, applications and markets.
Springer-Verlag.

Oskouei, R.J., Varzeghani, H.N., & Samadyar, Z. (2014). Intelligent

agents: A comprehensive survey. International Journal of Electronics
Communication and Computer Engineering, 5, 790798.
Oxman, S., & Wong, W. (2014). White paper: Adaptive learning systems.
DV X Innovations DeVry Education Group.
Popescu, E. (2008). Dynamic adaptive hypermedia systems for e-learning,
Universitede Technologie de Compiegne, Ph.D. thesis, HAL Archives.
Retrieved January 1, 2015, from https://tel.archives-ouvertes.fr/tel00343460
Radwan, N. (2014). An adaptive learning management system based on
learners learning style. International Arab Journal of e-Technology, 3,
228234.
Scutelnicu, Lin, F., Kinshuk, T.-C. L., Graf, S., & McGreal, R. 2014.
Integrating JADE agents into Moodle. Retrieved 1 Jun 2014, from
http://sgraf.athabascau.ca/publications/scutelnicu_etal_IAWES07.pdf