Collaborative Tutoring System Adaptive for Tutor's Learning Styles

based on Felder Silverman Model

Karima Boussaha

and Samia Drissi

Department of Computer Science, Research Laboratory on Computer Science’s Complex Systems (ReLa(CS)2),

Larbi ben M’hidi University, Oum El Bouaghi, Algeria

Department of Computer Science, Cherif Mssadia University, Souk Ahras, Algeria

Keywords: Tutors' Collaboration, CSCL, CSCC, Computer-Supported Collaborative Coaching, Tutoring, Collaboration,

CSCTT, Tutor, Tutor Group, Collaborative Tutoring, Coaching, Coach, Coach Group, Collaborative

Coaching, Personalized Learning, Learning Style, Learning Strategies.

Abstract: In the past decade, in the context of CSCL various personalization techniques have been proposed for

developing adaptive and collaborative e-learning systems, these later are specifically designed to assist and

support learners in their learning process. It is not only learners but also the tutors who experience

difficulties in the learning process. In particular, new recruits may not have enough experience to help their

learners. In this research paper, we investigate these ideas to propose a Computer-Supported Collaborative

Coaching System with Four-Dimensional Personalization Criteria based on Felder Silverman model called

CSCCS @ FDPC-FS. This system aims to create a virtual space based on the exchange of information and

experiences between experienced tutors in higher education institutions (coaches) to help new recruits

coaches who have difficulties and try to encourage, motivate, and provide them with needed experiences to

help them break out of isolation and use their solid information to guide their learners. This system offers

two strategies, to help new recruits coaches: either the first strategy offers to the tutor to acquire the

experience with a learning strategy of the basic notions of the tutoring process, this first strategy (learning)

combining and adapting teaching strategies, learning styles, and electronic media according to Felder-

Silverman's learning style model. And the second strategy offers to the tutors the possibility to collaborate

with other experienced colleagues, to gain the experience and the know-how. The collaboration strategy

offers a classification algorithm for forming coaches' groups, the forming groups algorithm base on two new

proposes profiles: collaborator and group profile.

1 INTRODUCTION

In the context of CSCL, the majority of CSCL

systems neglect the aspect of group formation by

grouping the learners randomly (Alfonseca et al.,

2006). Recently, many researches use many criteria

for grouping learners: we can cite among these

criteria: their profiles, personal information (age,

gender, class, etc.) (Analoui et al., 2013.), behaviors,

and knowledge, learning styles (Grigoriadou et al.,

2006), etc. Other works group learners using their

abilities and their thinking styles.

Many techniques were used to group learners,

Artificial intelligence and Bio-inspired techniques

are among the most common techniques. In the

context of bio-inspired techniques, Montazer and

Rezaei (2012) have introduced an optimization

approach in the e-learning field to improve the

grouping methods. Abnar and his colleagues (Abnar

et al., 2012) form learning groups by an iterative

process based on a genetic algorithm. Other

researchers discovered (Ghorbani et Montazer,

2012) that grouping learners by PSO (Particle

Swarm Optimization) technique based on their

cognitive styles improved the accuracy of grouping.

In another approach, Zedadra et al.(2016) presented

a new approach of learners grouping in collaborative

learning systems. This grouping process is based on

traces left by learners. The proposed approach

consists of two main algorithms: (1) the circular

grouping algorithm and (2) the dynamic grouping

algorithm (used to update groups). The authors'

proposed approach used the same behavior of

penguins' colony. So, we found many works about

200

Boussaha, K. and Drissi, S.

Collaborative Tutoring System Adaptive for Tutor’s Learning Styles based on Felder Silverman Model.

DOI: 10.5220/0010425602000207

In Proceedings of the 13th International Conference on Computer Supported Education (CSEDU 2021) - Volume 2, pages 200-207

ISBN: 978-989-758-502-9

grouping learners and learners collaboration. But,

collaboration among learners is not enough to solve

some problems like some learners find it difficult to

communicate and share experiences within the

group (Rojano-Cáceres et al.,2017). As a result, the

monitoring functionality is required in these

environments. Tutoring is a key element of any

distance learning system, which has been applied in

several fields. In the educational field, this task has

become indispensable, especially in higher

education institutions.

The primary objective of tutoring is to support

the learners throughout the learning so that it fully

reaches the educational objectives set by the

educational institution.

Furthermore, it aims at supporting all the

activities of learners and assisting them to find

learning difficulties and problems. In other words,

distance tutoring or e-tutoring is referred to all the

activities that support learners in their learning

process (Kopp et al., 2012; AbuEloun and Abu

Naser,2017; Benjamin D. Nye et al.,2018).

Learning style is a learner characteristic

indicating how a learner learns and likes to learn

(Keefe, 1991). For example, some learners prefer

graphical representations and remember best what

they see, others prefer audio materials and remember

best what they hear, while others prefer text and

remember best what they read. Some learners like to

be presented first with the definitions followed by

examples, while others prefer abstract concepts to be

first illustrated by a concrete, practical example.

In the few recent years, various personalization

techniques have been proposed for developing

adaptive e-learning systems, and have revealed the

benefit of such an approach. In this respect,

according to (Al-Azawei & Lundqvist, 2015) many

personalized or adaptive learning systems have been

developed focusing on a range of learner's personal

information, such as their profiles (e.g., gender, age,

knowledge level, and background data), learning

portfolios, and preferences. Recently, researchers

have largely focused on learning styles due to

several reasons. According to literature, learning

styles have widely been used to avoid a ‘one-size-

fits-all’ teaching approach (Al-Azawei & Badii,

2014; Felder& Brent, 2005).

There are many studies on the effectiveness of

combining multimedia and hypermedia with

learning styles in educational systems. They attempt

to associate specific e-media characteristics to

different categories of learners and propose

instruments and methods for assessing learning

style. Most of these studies are based on the Felder-

Silverman learning style model (FSLSM) (Felder

and Silverman, 1988). Examples of such systems

include CS383 TANGOW, and PHP Programming

Course (Hong & Kinshuk, 2004).

Learning strategies are the strategies used to

remember, learn and use information. In this regard,

some of the previous studies worth mentioning are

for example those of Dunn, who insists on the

importance of teaching the learners by using

methods that adapt to their conceptual preferences.

However, very few researchers give any idea of

the correspondence between electronic media and

the appropriate teaching and learning styles and very

few studies give an idea of the appropriate

combinations of electronic media and learning styles

that are more effective than others.

So, as we talk about learners' grouping, and

learners' learning styles, tutors, especially new

recruits also need to collaborate and learn according

to their learning styles (Indira, 2019; Tadjer et al.,

2018,2020)

Tutors need to work in groups in some cases for

example: when learners' needs do not belong to the

tutor's skills, the learner's queries will not be

satisfied, so in this case, the tutor must collaborate

with others to can help his /her learner. Also in the

case of new recruits, they have needed to collaborate

to get the experience from their experienced friends

(Indira, 2019; Tadjer et al., 2018,2020).

Tutors especially new recruits need to learn

about basic tutoring notions with their learning

styles to give them the advantage of autonomy by

gaining experience and avoiding the problems of

disorientation and cognition. The learning according

to the tutors' learning style offers to them a

navigation adaptation (Drissi & Amirat, 2017).

In the literature, the authors found some recent

IT(intelligent Tutoring ) platforms we can cite

among them: the work presented in AbuEloun and

Abu Naser(2017), Another work was presented by

Benjamin D. Nye et al (2018). Of course, these

works focused on tutoring in its classic concepts,

which means the tutoring process between student

/tutor without take into count neither the learning

according to the tutors' learning style neither the task

of collaboration between tutors in their

methodologies.

In this paper, we aim to propose a novel

Computer-Supported Collaborative Coaching

System with Four-Dimensional Personalization

Criteria based on the Felder-Silverman model called

CSCCS @ FDPC-FS. Our system presents a general

framework used to teach new tutors recruits

(coaches) the basic concepts about the tutoring

Collaborative Tutoring System Adaptive for Tutor’s Learning Styles based on Felder Silverman Model

201

process, in higher education institutions. This

proposed system combines with two strategies:

• Learning strategy: uses combining and

adapting teaching strategies, learning styles,

and electronic media According to Felder-

Silverman's learning style model. More

specifically, This strategy focuses on the

proposal for an adaptive taxonomy that will be

used to release the fourth levels of adaptation

which are: content level adaptation, link-level

adaptation, presentation level adaptation, and

collaboration level adaptation of an educational

collaborative tutoring system (Drissi &

Amirat,2017). While based on the four

dimensions of Felder-Silverman's learning style

model. This strategy is used when the new

tutor recruit (coach) chooses to learn the basic

tutoring notions alone.

• Collaboration strategy: uses the concepts of

collaboration and joining in different groups of

experienced tutors. This second strategy is used

when the new tutor (coach) prefers to get help

from his/her colleagues with prior.

This paper is organized as follows. In the second

section, we give a brief of the Felder-Silverman

learning style model. The architecture of a CSCC

system is presented in section three. We conclude

with a conclusion and future works.

2 FELDER SILVERMAN'S

MODEL

In this research, we are focusing on the Felder-

Silverman learning style model (FSLSM) because

• the (FSLSM) was widely used, more

specifically, in Technology Enhanced Learning

(TEL) (Al-Azawei & Badii, 2014; Drissi &

Amirat, 2017).

More specifically, according to Carver, Howard, &

Lane (1999), the Felder Model is most appropriate

for hypermedia courseware. So, in our case we

applied this model in our CSCCS @ FDPC-FS, to

offer to the new recruit to learn the basic notions of

the tutoring process according to his learning style.

So, automatically the tutor becomes a learner.

FSLSM contains four dimensions when the

learner is characterized by a specific preference for

each of these dimensions. Each dimension includes

two variables as shown in figure 1. As detailed in

(

Drissi & Amirat, 2017), the first dimension covers

sensing versus intuitive learning. Students who

enjoy studying from facts and concrete learning

materials are students who foster a sensitive learning

style. In contrast, intuitive learners are more

motivated by abstract learning, such as theories and

their underlying meanings. They can discover

possibilities and relationships and tend to be more

innovative and creative than sensing learners. The

second, visual-verbal dimension differentiates

learners who remember best and

who learn best

through vision (e.g., pictures, diagrams, and flow-

charts), and learners who benefit the most from

textual representations, whether written or spoken.

In the third dimension, the learners are rated

between an active and a reflective way of processing

information. Active learners tend to be more

interested in communication with others and prefer

to learn by working in groups. In contrast, reflective

learners favour individual work or perhaps prefer to

work in small groups together with one good friend.

Finally, the fourth dimension characterized learners

according to their understanding. Sequential learners

have linear learning progression. So, they prefer to

learn in small incremental steps. Whereas, global

learners use a holistic thinking process and learn in

large leaps. Table 1 details the description of Felder-

Silverman dimensions.

Figure 1: Felder-Silverman learning style model.

CSEDU 2021 - 13th International Conference on Computer Supported Education

202

Table 1: Description of Felder- Silverman dimensions (Drissi & Amirat, 2017).

Learning Style Description

Sensing

Prefer concrete facts, data, and relation to the real world around. Rather deal with facts, raw data, and

experiments, they are patient with details but do not like complications.

Intuitive

Focus on ideas and possibilities, Prefer abstraction, theories, and models. Rather deal with principles and

theories, are easily bored when presented with details, and tend to accept complications.

Sequential

Orderly, step by step, and sequential. Follow a linear reasoning process when solving problems and can

work with a specific material once they have comprehended it partially or superficially.

Global

See everything as a whole. Take big intuitive leaps with the information, may have a difficulty when

explaining how they got to a certain result, need an integral vision.

Visual Easy for them to remember what they see: images, diagrams, time tables, films, etc.

Verbal Remember what they have heard, read, or said.

Active

Motivated, prefer trial-and-error. Enjoy discussion rather than learning independently. “learning by

doing” describe how active students learn. Learn by working in groups and handling stuff.

Reflective

Learn better when they can think and reflect on the information presented to them. Learn a good deal

from independent work. Work better alone or with one more person at most.

“Learning by thinking” could describe Reflective students.

3 ARCHITECTURE OF THE

PROPOSED SYSTEM

Our CSCCS @ FDPC-FS system is organized in the

form of three basic components: models component,

collaboration component, and learning component.

These three components interacted to adapt different

aspects of the instructional process. Figure 2

illustrates the system architecture.

3.1 The Models Component

This module is divided into three sub-modules:

3.1.1 Tutor Model

In CSCCS @ FDPC-FS proposed, to model our tutor

(coach) we will follow two phases:

• Phase 1: In our approach, the tutor (coach)

can be modeled first by the typical

characteristics that are grouped in a facet

identification that contains personal data for

example username, password, unique ID, age,

sex, e-mail. These data are obtained using a

questionnaire that the tutor must complete on

their initial login.

• Phase 2: Selection of learning styles of tutors

is performed using the Index of Learning

Style Questionnaire (ILQ), developed by

Felder and Soloman (1997) which is used to

categorize the learners into four dimensions

(Sensing/Intuitive, sequential/Global,

Visual/Verbal, and Active/Reflective). The

description of these dimensions is detailed in

table 1 previously.

So in this phase, the tutor must complete a

questionnaire containing 44 questions (11 items per

dimension). Knowing that, each tutor has a personal

preference for each dimension. These preferences

are expressed with values between +11 to -11 per

dimension, with steps +/-2. These measures come

from the 11 questions that are posed for each

dimension. For example, when answering a

question, with an active preference, +1 is added to

the value of the active/reflective dimension whereas

an answer for a reflective preference decreases the

Collaborative Tutoring System Adaptive for Tutor’s Learning Styles based on Felder Silverman Model

203

Figure 2: Architecture of the proposed system CSCCS @ FDPC-FS.

value by 1. And for this, each question is answered

either with a value of +1 (answer a) or -1 (answer b).

while (Answer a) corresponds to the preference for

the first pole of each dimension (active, sensing,

visual, or sequential), and (answer b) to the second

pole of each dimension (reflective, intuitive, verbal,

or global).

3.1.2 Domain Model

In this domain model, the tutor (coach) found all the

needs of the basic notions of the tutoring process.

This domain is represented by three levels: course,

chapter and finally learning objects.

3.1.3 Evaluation Test

This module is responsible for offering some tests

for the tutors.

3.2 The Learning Component

If the new tutor (coach) prefers to learn alone all the

basic notions of the tutoring process, so he/she must

choose the learning strategy, this module is divided

into three sub-modules:

3.2.1 Adaptive Module

The adaptive module aims to provide a personalized

learning resource for tutors (coaches), especially

learning content by suggesting personalized learning

paths and adaptive layouts when we proposed an

adaptive taxonomy that integrates learning

strategies, learning styles, and electronic media.

This adaptive taxonomy focuses on a set of

resources summarized as navigation tools,

Collaboration and communication tools, overview

tool, and a set of learning objects.

TUTO

ILQ

Application

server

Tutor

Registration

Consultation of courses

Realization of evaluation test

Models

component

Collaboration

component

Learning

component

Communication

Interface

Editor of basic tutoring

Evaluation tests

Tutor model

Domain model

Ada

tive module

Evaluation

Collaborator

Grou

rofile

Forming Groups

module

Learning Module

CSEDU 2021 - 13th International Conference on Computer Supported Education

204

The learning objects are presented in the form of

text, picture sound, and animation. Furthermore, our

adaptive taxonomy is adapted to release the fourth

levels of adaptation: content level adaptation, link-

level adaptation, presentation level adaptation, and

collaboration level adaptation of an educational

hypermedia course, while based on the four

dimensions of Felder-Silverman's learning style

model.

3.2.2 Learning Module

The learning module based on the tutors' four

dimensions of Felder-Silverman's learning style

model to provide tutors (coaches) with personalized

learning content by suggesting personalized learning

paths and adaptive layouts

3.2.3 Evaluation Module

The evaluation module presents to the new tutors

(coaches) a set of evaluation tests to evaluate their

knowledge in two levels the pre-test level and the

post-test level. This module provides tutors

(coaches) with different types of exercises including

multiple-choice questions (MCQ) and matching

questions.

In CSCCS @ FDPC-FS, evaluation tests were

automatically generated using an editor of basic

tutoring notions called Hot potatoes that is a

software suite that includes five applications to

create exercises to upload on the web. In our system,

two applications are used: JQuiz editor (multiple-

choice questions or MCQs) and JMatch (Editor of

matching exercises).

3.3 Collaboration Component

If the new tutor (coach) prefers to learn all the basic

notions of the tutoring process, with the help of his

/her experienced tutors so, he/she must choose the

collaboration strategy.

The methodology proposed for creating the

tutors' groups is based on the set of tutor's (coaches)

profiles. Tutor's (coach's) profile is a tool to get an

idea about him/her, and with whom he /she interact.

The tutors'(coach's) profile is composed of

information that identifies the ability of the coach in

practice. There is a personal profile (name, age,

email..etc.), the cognitive profile (diplomas,

experience, interest domains…etc.), the behavioural

profile (teaching methods, social relationship,.. etc.).

When the coach (tutor) starts working within the

system, this latter has no prior information about his

collaboration and his groups. However, the system

could not give any preferences for him. Therefore,

the model of this tutor must have an efficient way of

inferring initial information about the tutor. The

proposed approach is taken into account by an

educational system. In this system, all the necessary

information is collected to build the profiles of this

actor. For obtaining his group and collaboration

profiles and knowledge level, a pre-test is used. The

tutor can choose one or more questions according to

his skills to do the task. The collaboration

component is divided into three sub-modules:

3.3.1 Collaborator Profile

This new profile has five possible values (very

passive, little passive, little active, active, highly

active). The collaborator profile aims to pre-

classify coaches according to their level of collabora

tion in the different activities of their colleagues.

We must note that we need this profile when a coach

has the desire to join a group.

3.3.2 Group Profile

The interest domain of the coach is related to his

specialty diploma we try to reduce the number of

groups to eight (e-learning - multi-agent system,

Artificial vision, image processing, artificial

intelligence, networks, information system, internet

of things).

3.3.3 Forming Groups Module

So, our idea is we want to classify a new coach

(tutor) who registers in the system to join a group of

coaches that already exists in the system.

The KNN (K Nearest Neighbors) algorithm makes it

possible to determine the k nearest neighbors, and

then, for example, to classify a data item in one

category or another. The classification process is

based on the Euclidean distance between the data

(Zhang,

2016).

In our case, we want to classify a new coach

(tutor) who registers in the system to join a group of

coaches (tutors) that already exists in the system.

The classification is made based on the two new

profiles proposed: the coach's prior group profile and

the coach's collaborator profile. That is to say, for a

new coach, the system calculates two coordinates:

the group performance and the collaboration value:

for the group performance will be calculated from

his prior group profile, and for the collaboration

value will be calculated from his collaborator

profile. Each coach C will be represented by a point

Collaborative Tutoring System Adaptive for Tutor’s Learning Styles based on Felder Silverman Model

205

whose first coordinate is the group performance and

the second coordinate is the collaboration value, and

we determined the number of K closest neighbors of

the other coaches registered in our system using the

Euclidean distance between the new coach and all

other already registered coaches.

4 CONCLUSION AND FUTURE

WORKS

In this paper, as the first step of our research project,

we have presented the design of a collaborative

tutoring system adaptive for tutor's learning styles.

The proposed system supports two strategies:

The Collaboration strategy: among human

coaches in higher education institutions, we describe

the scenarios of the collaboration between coaches

by proposing two new profiles (the coach's

collaborator profile, the group profile) using for

forming coaches' groups these profiles are used by

the classification algorithm for forming groups. This

strategy gives the new recruit tutor the benefit of

gaining experience and the know-how from his more

experienced colleagues.

The learning strategy: for combining and

adapting teaching strategies, learning styles, and

electronic media according to Felder-Silverman’s

learning style model. More specifically, in this

paper, we have proposed an adaptive taxonomy used

to release the fourth level of adaptation by using

firstly, the «perception dimension» of Felder-

Silverman’s model to adapt learning content.

Secondly, the «understanding dimension» to realize

the navigation level adaptation. Thirdly, the «entry

Chanel dimension» to realize the presentation level

adaptation. And finally, the «processing dimension»

to realize the collaboration level adaptation. This

strategy gives the new recruit tutor the advantage of

autonomy by gaining experience and avoiding

problems of disorientation and cognitive overload

since it offers a navigation adaptation according to

the tutor's learning style.

As future work, we plan to extend the proposed

approach by developing a prototype of the

Computer-Supported Collaborative Coaching system

with Four-Dimensional Personalization Criteria

based on Felder Silverman model called CSCCS @

FDPC-FS, also the development of the Classification

algorithm for forming coaches' (tutors') groups.

REFERENCES

Abnar, S., Orooji, F., Taghiyareh, F., 2012. An

evolutionary algorithm for forming mixed groups of

learners in web-based collaborative learning

environments, ICTEE.2012, Proceedings of the IEEE

International Conference on Technology Enhanced

Education, pages 1-6. IEEE.

DOI:10.1109/ICTEE.2012.6208612.

AbuEloun, Nour, N., & Abu Naser, Samy, S. (2017).

Mathematics intelligent tutoring system, International

Journal of Advanced Scientific Research, 2(1)(2017),.

11-16.

Al-Azawei, A., & Badii, A. 2014. State of the art of

Learning Style Based Adaptive Educational

Hypermedia Systems (LS-BAEHSS), International

Journal of Computer Science & Information

Technology, 6(3), 1-19.

Al-Azawei, A., & Lundqvist, K., 2015. Which Factors

Affect Learner Achievement? Analysing the Role of

Psychological, Surface Level, Environmental and

learner Effort Variables. International Journal of

Education, 7 (2), 215-231.

Alfonseca, E., Carro, R. M., Martín, E., Ortigosa, A.,

Paredes, P., 2006. The impact of learning styles on

student grouping for collaborative learning: A case

study. User Modeling and User-Adapted Interaction,

16(3- 4), 377–401. DOI:10.1007/s11257-006-9012-7.

Analoui, B. D., Sambrook, S., Doloriert, C. H., 2013.

Engaging students in group work to maximize tacit

knowledge sharing and use. The International Journal

of Management Education., 12(1), 35–43.

DOI:10.1016/j.ijme.2013.08.002.

Benjamin, D., N., Philip I., Pavlik Jr., Windsor, A., Olney,

A., Hajeer, M., & Hu, X., (2018). SKOPE-IT

(Shareable Knowledge Objects as Portable Intelligent

Tutors): overlaying natural language tutoring on an

adaptive learning system for mathematics,

International Journal of STEM Education 5:12

https://doi.org/10.1186/s40594-018-0109-4.

Carver, C. A., Howard, R. A., & Lane, W. D. 1999.

Enhancing Student Learning Through Hypermedia

Courseware and Incorporation of Student Learning

Styles. IEEE Transactions on Education, 2(3), 33-38.

Drissi, S., & Amirat, A. (2017) Adaptation with Four

Dimensional Personalization Criteria Based on Felder

Silverman Model. Int. J. Distance Educ. Technol.

15(4): 1-20 (2017).

Felder, R., & Brent,R., 2005. Understanding student

differences. J. Eng. Educ., 94(1), 57–72.

Felder, R. M., & Soloman, B. A., 1997. Index of Learning

Styles questionnaire.

Felder, R. M., & Silverman, L. K., 1988. Learning and

teaching styles in engineering education. Engineering

Education, 78 (7).

Hong, H., & Kinshuk., 2004. Adaptation to Student

Learning Styles in Web Based Educational Systems.

Proceedings of EDMEDIA 2004, Chesapeake, VA:

Ghorbani, F, Montazer, G. A., 2012. Learners grouping

improvement in e-learning environment using fuzzy

CSEDU 2021 - 13th International Conference on Computer Supported Education

206

inspired PSO method. Proceedings of the Third

International Conference on E-Learning and E-

Teaching (ICELET) pages 65-70, IEEE.

DOI:10.1109/ICELET.2012.6333367.

Grigoriadou, M, Papanikolaou, K., Gouli, E., 2006.

Investigating how to group students based on their

learning styles. Proceedings of theSixth International

Conference on Advanced Learning Technologies

pages 1139-1140.

Indira, B., Valarmathi, K., & Devaraj, D. (2019). An

approach to enhance packet classification performance

of software-defined networks using deep learning. Soft

Computing, 23(18), 8609–8619.

Keefe, J. W., 1991. Learning style: Cognitive and thinking

skills. Reston, VA: National Association of Secondary

School Principals.

Kopp, B., Matteucci, M., C., Tomasetto, C., 2012. Tutorial

support for collaborative online learning: An

explorative study on experienced and inexperienced e-

tutors. Computers & Education, 58(1).

Montazer, G. A., & Rezaei, M. S. (2012). A New

Approach in E-Learners Grouping Using Hybrid

Clustering Method. Proceedings of the International

Conference on Education and e-Learning Innovations

(pp. 1-5). IEEE.

Rojano-Cáceres, J., R., Quintana, F., R., &Guerrero, E., B.

(2017). Assessing collaboration from real computer-

supported collaborative learning sessions. In

Proceedings of the XVIII International Conference on

Human-Computer Interaction (Interacción ’17). 1–8.

Tadjer, H., Lafifi, Y., & Seridi-Bouchelaghem, H. (2018).

A new approach for assessing Learners in an Online

problem-based learning environment. International

Journal of Information and Communication

Technology Education (IJICTE), 14(4), 18–33.

Tadjer, H., Lafifi, Y., Seridi-Bouchelaghem, H., &

Gülseçen, S.,(2020) Improving soft skills based on

students’ traces in problem-based learning

environments Interactive Learning Environments,

Zedadra, A., Lafifi, Y., Zedadra, O. (2016). Dynamic

Group Formation based on a Natural Phenomenon.

International Journal of Distance Education

Technologies 14 (4). October-December 2016.

Zhang, Z (2016) .Introduction to machine learning: K-

nearest neighbors, Annals of Translational Medicine

4(11):218-218.

Collaborative Tutoring System Adaptive for Tutor’s Learning Styles based on Felder Silverman Model

207