Ontology-Driven Intelligent Group Pairing in Project-Based

Collaborative Learning

Asma Hadyaoui

and Lilia Cheniti-Belcadhi

Sousse University, ISITC, PRINCE Research Laboratory, Hammam Sousse, Tunisia

Keywords: Assessment, PBCL, Peer-Group Feedback, Ontology, PAPI Learner, LOM, Intelligent Pairing,

Agglomerative Clustering, Critical Thinking, Creativity.

Abstract: In this research project, we investigate the influence of real-time online feedback from peer groups on the

assessment of group work in the setting of Project-Based Collaborative Learning (PBCL). Peer feedback plays

a crucial role in assisting students in evaluating their learning progress and acquiring valuable skills.

Nevertheless, its effectiveness in group environments has yet to be explored. To tackle this issue, we propose

an intelligent approach driven by ontologies to collect pertinent peer group feedback from the most compatible

groups. We make use of agglomerative clustering to identify groups that closely match and connect them to

exchange feedback. We utilize the information embedded in the ontology to create pairs of groups exhibiting

similar behaviors and dynamics during project-based learning activities. To assess the effectiveness of our

approach, we divide our dataset into two equal parts. We apply our intelligent pairing method to one half and

a random approach to the other. We conduct assessments both before and after peer group feedback to measure

its impact on project outcomes, including critical thinking and creativity. The results indicate a substantial

improvement in project outcomes, particularly in terms of critical thinking and creativity, due to peer group

feedback. Additionally, the groups formed using the agglomerative clustering algorithm demonstrate a higher

increase in project validation (8.33%) compared to the random approach (5.27%). Our research underscores

the effectiveness of integrating intelligence into the peer group feedback process, especially in the context of

PBCL. The proposed ontology presents a promising solution for optimizing the assessment process, leading

to improved results and the cultivation of critical thinking and creativity among students.

1 INTRODUCTION

Utilizing formative assessment has demonstrated its

potential to enhance student engagement and promote

knowledge retention. However, it's crucial to

acknowledge the existing methodological challenges

when it comes to substantiating claims of its

effectiveness (Bennett & Service, 2014). Formative

feedback holds a pivotal role in the realm of higher

education assessment. Hence, it becomes imperative

to explore alternative assessment approaches that

incorporate and enhance formative feedback.

Feedback is widely recognized as a cornerstone of

educational practice and a fundamental catalyst for

students' development and learning (Clark, 2012). At

the heart of the learning process lies feedback, as it

provides students with the necessary skills to

https://orcid.org/ 0000-0002-7006-8735

https://orcid.org/ 0000-0001-8142-6457

construct meaning and independently regulate their

educational path. This encompassing concept

encompasses both structured, formalized feedback

and comprehensive feedback, characterized by in-

depth, open-ended commentary (Hwang et al., 2023).

Peer feedback emerges as a pedagogical practice that

empowers students to engage in reciprocal

assessment, allowing them to provide and receive

constructive insights on their peers' work concerning

the same assignment, thereby elevating their

academic achievements. Peer-to-peer feedback

constitutes an educational approach that proves

highly advantageous for students (Topping, 1998).

Through this process, students not only enhance their

comprehension of learning objectives and success

criteria but also cultivate their ability to provide

valuable feedback to their peers. Collaborative

152

Hadyaoui, A. and Cheniti-Belcadhi, L.

Ontology-Driven Intelligent Group Pairing in Project-Based Collaborative Learning.

DOI: 10.5220/0012237500003584

In Proceedings of the 19th International Conference on Web Information Systems and Technologies (WEBIST 2023), pages 152-163

ISBN: 978-989-758-672-9; ISSN: 2184-3252

learning settings, featuring peer feedback, yield

advantages not only for the receiver but also for the

giver of feedback. The act of aligning learning

objectives and success criteria within another's work

provides invaluable learning experiences. This

assertion is corroborated by a study (Dong et al.,

2023) underscoring the importance of considering

students' competencies and attitudes in the dynamics

of giving and receiving peer feedback. Furthermore,

this research enhances our comprehension of peer

feedback literacy and illuminates the complex

relationship between feedback literacy and student

attributes, as substantiated by the creation of a

comprehensive measurement instrument. In a similar

vein, another study (Wu & Schunn, 2023) has posited

that encouraging students to actively engage in

activities that necessitate explanations and revisions

in response to feedback can yield significant

improvements in their learning outcomes. Over the

past few decades, there has been considerable

academic interest in peer feedback, with educators

advocating for its integration into educational

practices. One notable study by (Lin & Lin, 2017)

explored the effects of anonymous online peer

assessment facilitated through a Facebook

application on participants' perspectives regarding

learning, fairness, and attitudes. The results indicated

that the utilization of online anonymous peer

assessment had a favorable effect on participants'

attitudes and their perception of the learning

experience. However, another study (Banister, 2020)

explored the challenges and attitudes associated with

peer feedback among undergraduates studying

academic and business English. By employing the

principles of Exploratory Practice, the authors aimed

to understand why students may not value peer

feedback. Additionally, in a previous study (Cheniti

Belcadhi, 2016), we proposed an intelligent

framework for tailored feedback, leveraging

Semantic Web technologies. This framework

provides individualized feedback for self-assessment

and proves valuable in the context of lifelong

learning. In a different setting, the study (Chang et al.,

2020) gave fifth-grade students a chance to learn

about a geological park in a natural science class by

using peer review in virtual reality design activities.

However, while peer assessment in collaborative

learning environments typically involves interactions

between individual learners in a group, there is a lack

of research addressing peer assessment between

groups of learners, where both the assessee and the

assessor are groups. Given these considerations, we

recommend a reevaluation of the peer group feedback

procedures by incorporating real-time feedback

mechanisms to foster effective peer-to-peer

communication and ensure clear performance

assessment. Drawing on a comprehensive review of

existing research, our proposal centers on peer

feedback within a PBCL context. It presents a

conducive environment for students to assess each

other's performance, offer scores or grades, and

provide constructive written or oral feedback to

enhance learning and growth (Hwang et al., 2023).

Our investigation will address the following research

inquiries: Does peer feedback impact group

performance? How can we determine the most

suitable peer group for both receiving and providing

feedback? Does the selection of peers influence the

quality of feedback and, consequently, its

effectiveness?To address our research queries, we

carried out an investigation involving a cohort of 312

undergraduate students who were enrolled in the

initial stage of their degree program within our

academic department. The study was conducted

during the first semester of the academic year 2022-

2023, with students being organized into smaller

groups, comprising either 4 or 3 learners. Each group

was assigned the same collaboration project. We

proposed an intelligent method for generating pairs of

groups for them to provide feedback, taking into

account their ontological attributes and

characteristics. We designed the ontology called

PeerGroupOnto, to capture the key concepts and

relationships relevant to peer group formation and

feedback in the context of PBCL. The proposed

ontology leverages semantic web technologies and

eLearning standards to ensure data reusability and

interoperability. Our proposed intelligent approach

relies also on unsupervised machine learning, namely

the Agglomerative clustering algorithm as it allows

us to construct a tree-like structure from groups of

comparable students. We are delighted with the initial

grouping since we desire to form couples. As a

grouping criterion, we used the similarity of intra-

group interactions, which reflects group dynamics.

To evaluate our method, we divided our dataset in

half and applied our proposed method solely to one

half, while the other half was randomly paired. An

evaluation is conducted before and following peer

group feedback during a chat session. Through

synchronous discussions with peers, groups could

reevaluate their answers and project success

considering the many viewpoints supplied by their

peers.

The integration of ontology and the utilization of

an intelligent pairing approach based on the

agglomerative algorithm have demonstrated

significant synergies, leading to more effective and

Ontology-Driven Intelligent Group Pairing in Project-Based Collaborative Learning

153

meaningful collaborative learning experiences. The

ontology serves as a foundational knowledge

representation framework, capturing domain-specific

concepts, relationships, and properties relevant to the

assessment process. By formalizing these elements,

the ontology enables a structured and interconnected

learning environment, facilitating the seamless

identification and organization of peer groups.

In this paper, we start by delving into the relevant

literature surrounding collaborative learning settings.

This involves exploring how ontology and semantic

web technologies are applied in educational contexts.

Our focus then shifts to introducing our proposed

ontology in detail. This serves as the basis for

modeling collaborative eLearning environments and

facilitating the formation of peer groups, as well as

the assessment process. The integration of ontology

and semantic web technologies in our research

strengthens its foundation and contributes to the

advancement of collaborative learning practices.

Moving forward, we explain the methodology we

used to form peer groups and analyze the impact of

peer group feedback on project assessment outcomes.

Throughout our study's methodology description, we

walk through the steps we followed to arrive at our

conclusions. We thoroughly discuss these

procedures, draw conclusions, and finally wrap up

our investigation.

2 LITERATURE REVIEW

2.1 Collaborative Learning

Collaborative learning is an educational approach that

involves the gathering of students to engage in

discussions about a subject matter relevant to the

course or curriculum, to collectively address a

problem associated with the topic or generate a

product that is connected to the topic. By employing

this approach, students have the opportunity to

cultivate proficiencies that are pertinent to the

contemporary era, encompassing effective

communication, analytical reasoning, sound

judgment, effective leadership, and conflict

resolution (Kalmar et al., 2022). Numerous studies

have been undertaken within this setting. The study

(Chorfi et al., 2022) introduced a novel groupware

system based on Computer-Supported Collaborative

Learning (CSCL) to facilitate PBCL within a

computer programming education context. A study

conducted by the authors (Mhlongo et al., 2020)

revealed that PBCL has a positive influence on

students' overall academic performance and skill

development, ultimately boosting their self-

confidence. PBCL is particularly effective at

overcoming barriers to strategic knowledge, such as

groupthink and collaboration. Additionally, the

approach of a semester-long undertaking makes

solving complex problems easier and reduces mental

strain (Wang & Hwang, 2017). Also, prior research

(Hadyaoui & Cheniti-Belcadhi, 2023) centered on

predicting disengagement among learner groups in a

PBCL context. We created a Collaborative

Assessment Analytics Framework (CAAF) utilizing

ontologies and the accumulation of formative

assessment data. In another study (Awuor et al.,

2022), the researchers investigated the correlation

between students' proficiency in teamwork and their

level of satisfaction in a synchronous online flipped

group project-based course. The study specifically

focused on exploring the potential moderating

influences of group collective efficacy and flipped

learning. In the context of asynchronous Computer-

Supported Collaborative Learning (CSCL), a study

conducted by (Chen & Du, 2022) identified distinct

profiles of regulators by analyzing online indicators

of collaborative learners' utilization of individual-

oriented and socially shared metacognitive regulation

strategies. Furthermore, the researchers examined the

correlation between regulation profiles and the post-

CSCL conceptual understanding, motivation, and

learning self-efficacy of the students. The authors

(Aranzabal Maiztegi et al., 2022) presented a

methodology for creating well-balanced teams by

utilizing Belbin's roles. This approach aims to

enhance positive interdependence and individual

accountability among team members, ultimately

leading to improved team performance within a

project-based learning setting. The evaluation of the

student's performance has been conducted by

assessing the scores obtained throughout the project.

A significant performance advantage was observed in

the Belbin teams when compared to the self-selected

teams. Furthermore, the feedback, experiences, and

opinions of the students were gathered and compiled.

After the academic term, students who engaged in

Belbin teams conveyed a more favorable attendance

record, necessitating reduced extracurricular study

time, and displaying an elevated level of enthusiasm

towards the subject matter. In a previous study

(Hadyaoui & Cheniti-Belcadhi, 2022b), an

assessment ePortfolio-based recommender system for

individualized formative assessment in a

collaborative online learning environment was

developed. This action was taken to improve student

performance. The learner's ePortfolio was linked to

the ePortfolios of other learners using the same

WEBIST 2023 - 19th International Conference on Web Information Systems and Technologies

154

assessment platform, allowing for the generation of

beneficial recommendations. In addition, a separate

study (Ismail et al., 2023) investigated the efficacy of

the think-pair-share (TPS) cooperative learning

technique in enhancing student engagement and

performance in number-building activities.

According to the findings of the study, TPS

cooperative learning considerably enhanced student

performance and motivation, while simultaneously

fostering creativity, teamwork, active learning, and

motivation. To accomplish our research objective, we

devote the majority of our research efforts to

determining the usefulness of PBCL-based peer

group input.

2.2 Semantic Web and Ontology

The emergence of the Semantic Web and the

incorporation of ontology have had a profound effect

on collaborative eLearning. By facilitating the

structured representation and organization of

knowledge, ontology-based semantic web

technologies play a vital role in collaborative

eLearning. Ontology facilitates the construction of

interconnected learning environments by formalizing

domain-specific concepts, relationships, and

properties. The authors (Murillo Zamorano et al.,

2019) created an ontology-driven collaborative

eLearning platform that effectively organized

learning materials, resulting in enhanced learner

comprehension and engagement. In the same context,

the study (Senthil Kumaran & Latha, 2023) proposed

an innovative method for providing adaptive access

to digital library learning resources. The method

employs an ontology-based multi-attribute

collaborative filtration system to enhance the

precision and efficacy of predicting and

recommending personalized learning resources.

Moreover, the Semantic Web makes a substantial

contribution to collaborative content discovery in

eLearning systems. Through the incorporation of

semantic annotations, learning resources become

more understandable to both human and machine

users. In a previous publication (García-González et

al., 2017), a novel educational tool was showcased,

employing Semantic Web technologies to enrich

lesson materials. Furthermore, an independent

research endeavor introduced a content recommender

system driven by ontology, to tackle cold-start

problems encountered by new users. Within this

recommendation model, ontology serves to describe

both the attributes of the learner and the

characteristics of the learning objects. Moreover,

Semantic Web ontology promotes the interoperability

and reusability of learning resources across a variety

of collaborative eLearning platforms. With

standardized metadata representation, educational

content can be readily integrated into a variety of

learning environments and shared with others. By

employing ontologies and eLearning standards,

including CMI5 for assessing the assessment path and

IEEE PAPI for the learner model, (Hadyaoui &

Cheniti-Belcadhi, 2022a) introduced a semantic web

approach rooted in semantic web principles. These

technologies ensure data reusability and

interoperability, ultimately improving the efficiency

and effectiveness of the personalized formative

assessment system within a collaborative learning

environment. In addition, Semantic Web ontology

enables personalized learning by leveraging learner

profiles and preferences to customize content

delivery. By utilizing ontology to capture individual

learning objectives and preferences, educators can

provide students with content that meets their specific

requirements. In (Joy et al., 2021), the research

introduced an ontology-driven semantic framework

to tackle the pure cold-start problem in content

recommenders. The ontology effectively captures

domain-specific details related to learners and

Learning Objects (LOs). Through SPARQL queries,

the semantic model leverages learner parameters,

including learning style, knowledge level, and

preexisting knowledge, to create natural learner

groups based on the learner ontology. In the context

of collaborative eLearning, semantic annotations play

a crucial role in enhancing contextual collaboration.

By enriching learning resources and collaborative

interactions with meaningful metadata, semantic

annotations offer relevant context and enhance

knowledge transfer. Additionally, previous research

(Nouira et al., 2018), in which the authors proposed

an architecture for assessment analytics that employs

semantic web technologies. They concentrated on

analyzing the collection of assessment activities,

assessment outcomes, and assessment context.

3 THE PEERGROUPONTO

DESCRIPTION

At the core of our approach lies the creation of a

specialized ontology that we call

PEERGROUPONTO, specifically designed for

PBCL. This ontology will act as a structured

representation of the knowledge domain, capturing

crucial concepts, relationships, and properties that

Ontology-Driven Intelligent Group Pairing in Project-Based Collaborative Learning

155

form the foundation of PBCL and peer group

feedback.

To ensure seamless integration with existing

eLearning systems and enhance interoperability, our

proposed ontology will embrace the widely

recognized IEEE Learning Object Metadata (LOM)

standard (“IEEE Standard for Learning Object

Metadata,” 2020). By adopting this standard, we

enable effortless exchange of learning resource

metadata, enabling educational materials to be

discovered and reused across diverse platforms.

The PeerGroupOnto, depicted in Figure 1,

constitutes a comprehensive set of essential classes

for our intelligent Semantic Web framework. These

Figure 1: PEERGROUPONTO's classes.

classes include Project-Based Learning,

Collaborative Learning, Peer Feedback, Intelligent

Group Pairing, Learning Outcomes, Assessment

Criteria, Learning Resources, Learners, Instructors,

and Educational Organizations. To ensure a seamless

implementation of our intelligent pairing peer group

feedback system in the PBCL context, each class is

enriched with relevant object and data properties,

establishing clear connections and attributes. This

well-defined ontology structure facilitates effective

communication and enables the successful

integration of our approach.

In the following, we will describe the essential

classes that constitute the ontology, known as the

PeerGroupOnto, and their respective properties,

which form the foundation of our intelligent Semantic

Web framework for peer group feedback in the PBCL

context.

Table 1: Learner-related Classes.

Class Name Description

Learner

Represents and captures

various aspects of a

learner's profile and

characteristics in e-

learning.

Performance

Focuses on the learner's

academic performance,

tracking their progress,

achievements, and

metrics.

PersonalInfo

Captures the learner's

personal information,

including name, gender,

age, contact details, etc.

PortfolioInfo

Pertains to the learner's

portfolio, showcasing

work, projects, and

achievements to showcase

skills.

Preferences

Records the learner's

learning styles, interests,

and preferences for

specific subjects or topics.

Relations

Deals with relationships

and connections between

learners and other entities

within the educational

context.

Security

Ensures appropriate

security measures and

access controls to protect

learners' sensitive

information.

WEBIST 2023 - 19th International Conference on Web Information Systems and Technologies

156

Table 2: Collaborative learning classes.

Class Name Description

Project-Based

Learning

Represents the learning approach where

students collaborate on Python

programming projects.

Collaborative

Learning

Represents the mode of learning where

students work together on shared tasks

and projects.

Peer Feedback Represents the feedback provided by

peers in a learning group, supporting

evaluation and improvement.

Intelligent

Group Pairing

Represents the methodology for

forming groups based on factors like

student expertise and skills.

Learning

Outcomes

Represents the expected knowledge,

skills, and competencies students

should gain from collaborative learning.

Assessment

Criteria

Represents the criteria and rubrics used

to evaluate the quality and performance

of Python projects.

Learning

Resources

Represents educational materials like

tutorials and code examples used to

support the learning process.

Learners/

Students

Represents individuals participating in

project-based collaborative learning.

Learning

Object (LOM

standard)

Represents a learning resource

described using the IEEE LOM

standard attributes.

Table 3: Object properties.

Object Property Description

hasLearningOutcome Connects a learning activity or

project to the intended learning

outcomes it aims to achieve.

hasAssessmentCriteria Links a project or assignment to

the criteria used for assessing it.

hasLearningResource Associates a learning activity

with the resources used to

support it.

involvesStudent Relates a project or collaborative

learning activity to the

participating students/learners.

involvesInstructor Relates a project or learning

activity to the

instructors/educators involved.

belongsToInstitution Connects a learning activity or

project to the educational

institution it belongs to.

providesPeerFeedback Links a project or activity to the

peer feedback provided by other

students.

hasFeedbackCriteria Associates peer feedback with

the criteria used for evaluating

it.

isRelatedToLearning

Object

Links a learning resource to the

corresponding LOM standard

attributes for metadata

representation.

Table 4: Data properties.

Data Property Description

feedbackContent Represents the content or text of

the peer feedback provided.

feedbackDate Represents the date when the

feedback was given.

feedbackRating Represents the rating or score

associated with the feedback.

groupSize Represents the size of the

collaborative learning group.

learningObjectTitle Represents the title of the

learning resource described

using LOM standard attributes.

learningObjectDesc

ription

Represents the description of the

learning resource.

learningObjectKey

words

Represents keywords or tags

associated with the learning

resource.

learningObjectLan

guage

Represents the language in

which the learning resource is

presented.

learningObjectFor

mat

Represents the format or media

type of the learning resource.

learningObjectLoca

tion

Represents the location or URL

where the learning resource can

be accessed.

learningObjectTech

nicalRequirement

Represents any technical

requirements needed to access

the learning resource.

learningObjectEdu

cationalLevel

Represents the educational level

or target audience for the

learning resource.

learningObjectInte

ndedEndUserRole

Represents the intended end-user

role for the learning resource.

4 PEER FEEDBACK

EXPERIMENT

Assessment is an essential component of any

teaching-learning process. It is the sole method for

acquiring a deeper understanding of the learners'

knowledge and skills, particularly their creative and

critical thinking, in a PBCL setting. In definitions,

these terms are occasionally interchangeable. In

actuality, they have different conceptualizations

because the consequences of human conduct vary. In

addition, one of today's requirements is that

individuals be able to handle everyday circumstances

with both skills (Birgili, 2015). We utilized a 0 to 10

scale to evaluate each learned skill during the

duration of the project. As the sum of the two

preceding grades, the final grade for the project is

based on a scale from 0 to 20. If the final score is 10

or greater, the project will be validated; otherwise, it

will be deemed invalid.

Ontology-Driven Intelligent Group Pairing in Project-Based Collaborative Learning

157

Our study involved 312 students enrolled in the

transportation and technology engineering first-

degree program at our faculty during the first

semester of the 2022-2023 academic year. The

course, titled "Programming Fundamentals," grouped

students into 60 teams of four and 24 teams of three

based on their scores. Initially, all groups exhibited

varying levels of proficiency in programming. After

eight weeks of classes, participants were introduced

to the educational content, which encompassed the

fundamental concepts of Python programming.

Participants worked together to develop a Python

application to address a problem. Students designed

and coded the application from the ground up, with a

focus on problem-solving. In online groups

consisting of 3 to 4 members, students engaged in a

collaborative process that encompassed

brainstorming ideas, analyzing problem statements,

and ultimately choosing one problem for their project.

The paramount objective was to identify a problem

that not only aligned with their educational goals but

also fostered meaningful collaboration, critical

thinking, and efficient communication. Once the

problem was defined, the next step involved

categorizing shared and subtasks. These subtasks

revolved around various aspects of their Python

application, including data processing, algorithm

implementation, user interface development, and

program logic. The allocation of subtasks was based

on the nature of the problem and the intended level of

complexity. Collaboratively, the team members

assumed responsibility for different subtasks, each

focusing on programming their respective portions.

To streamline the coordination and integration of

code contributions, the group employed Git, which

served as a version control system. Throughout this

process, the group maintained active communication,

sharing ideas, providing feedback, and offering

support in coding. Testing and debugging were

pivotal phases of the project to ensure its success.

Teams rigorously tested their work and addressed any

issues that arose. This iterative approach significantly

bolstered the software's engineering, functionality,

and reliability. The documentation and presentation

of the project held great importance. The groups

meticulously annotated their code and produced a

comprehensive project report. During their

presentations, they elucidated the problem statement,

outlined their proposed solution, and highlighted the

key features of their program. Effective coordination

and communication played a vital role throughout the

project's lifecycle. The groups utilized online

discussion forums, and chat rooms to communicate,

share progress reports, seek advice, and offer

feedback. This collaborative environment fostered

shared learning, the exchange of knowledge, and

problem-solving. For all that, we used the online

platform Moodle (Modular Object-Oriented Dynamic

Learning Environment) (Jan et al., 2018). Throughout

the experiment, we provided support and guidance to

the students whenever necessary, helping them

overcome any obstacles they encountered. The

collected data comprised the collaborative project

files submitted by the groups, as well as logs

documenting the authoring process and the online

activity of the forum groups.

Following the experiment, we extensively

evaluated the project's achievements and assessed the

contributions made in the discussion forum. The

primary experiment extended over six weeks in the

initial semester, and data from both the achievement

projects and forum discussions were subjected to

preprocessing for assessment purposes.

4.1 Collaborative Project Assessment

Table 5 provides a means to compare the initial

results with the updated findings, focusing on the

skills to be acquired through the project's activities.

Table 5: Skills to gain through the project’s activities.

Skills Approaches Project’s skills

Critical

thinking

Effectively

employs

logic and

reason when

relevant to

the context.

Development of a Python

program to address the

proposed issue:

 Creation of a list.

 Addition of new

entries to the

listModification of

list items.

 Sorting the list based

on criteria such as

Age, Baccalaureate

average, and Name.

Creativity Integrates

current

information

and

resources to

produce

fresh and

practical

ideas.

 Usage of menus to

display various

program functions.

 Utilization of

QtDesigner to design

visually appealing

interfaces.

WEBIST 2023 - 19th International Conference on Web Information Systems and Technologies

158

4.2 Peer Groups Forming

In our experiment, we divided our groups into two

sets consisting of 42 groups each, referred to as C1

and C2 in this study. For C1, group partners were

randomly paired. In contrast, for C2, we employed an

unsupervised machine learning approach called

Hierarchical Agglomerative Clustering (HAC). It is

widely recognized as one of the most popular

algorithms in unsupervised learning (Ben-david,

2014). The first step of HAC involves separating each

group of learners into individual clusters, resulting in

an initial number of clusters equal to the total number

of groups. Subsequent steps involve merging

comparable groups until all cases are grouped into a

single cluster. This process generates a dendrogram,

which is a tree-like structure representing the

relationships between clusters, as illustrated in

Figure 2.

Figure 2: Dendrogram Depicting the Results of Cluster.

Given our objective of creating peer pairs, we

stopped at the first step or level of grouping. The

success of HAC is heavily influenced by the selection

of a suitable linkage criterion, which measures the

similarity between two clusters (Ramos et al., 2021).

In our approach, we determined that the group's

contributions on the forum, reflecting their behavior

during the collaborative project, would serve as the

linkage criterion. Once the pairs were established,

members of each group participated in a chat session,

where they presented their work and responded to

questions from their peers, reciprocally.

5 RESULTS

This section describes the outcomes of our

experiment. We denote by C1 the category of groups

whose clustering pairs were produced at random. C2

is our grouping method based on the agglomerative

clustering technique depending on the intra-group

interactions (Total group contributions on the forum,

and the total times online during the project

execution). Then, we compare the assessment results

of the groups before and following peer group

feedback.

5.1 Comparing Project Validation

before and after Peer Group

Feedback for the Two Paring

Groups’ Approaches

As demonstrated in Table 6, the number of projects

that were validated before peer group assessment for

the two clustering categories (C1 and C2) has grown

after receiving feedback from their peers. Indeed, we

notice a rise in the rate of validation (expressed by 1).

The results demonstrate an increase of 5.27% in the

percentage of project validation from 73.68 to

78.95% for the C1 approach. For method C2, the

percentage of validated projects increased by 8.33%,

from 81.15 to 89.47%.

Table 6: Comparison of the average of project validation

before and after peer group feedback.

Project result before

peer group feedback

Project result after

peer group feedback

5.2 Comparing Project Learning Skills

Outcomes before and after Peer

Group Feedback for the Two

Paring Groups’ Approaches

Following are the outcomes of the group assessments

for each of the adopted strategies. According to the

subsequent tables, we utilized boxplots to compare

the findings acquired before and after peer group

feedback. Boxplots are a graphical representation that

Ontology-Driven Intelligent Group Pairing in Project-Based Collaborative Learning

159

effectively illustrates the distribution of numerical

data and provides insights into the presence of

skewness. This is achieved by displaying the quartiles

and the mean of the data. Boxplots display the five-

number summary, which encompasses the minimum,

maximum, first quartile, third quartile, and median

values, of a given dataset.

• Projects’ final score before and after peer

group feedback: As shown in Table 7, for the

initial method C1, 75% of the scores fall below

the upper quartile value 17. Thus, 25% of data

are above this value. However, after peer group

feedback, 25% of data are above 18.25 (a 1.25-

point increase). For the second approach C2,

75% of the scores fall below the upper quartile

value of 17.75. However, after peer group

feedback, 25% of data are above 19 (a 1.25-

point increase).

Table 7: Comparing the project’s final score.

• Projects’ creativity outcome before and

after peer group feedback: By table 8, for the

initial method C1, half the scores are greater

than 7 and a half are less. After the peer group

feedback, this value rose to 8 (a 1-point

increase). For the C2 approach, half the scores

are greater than 7 and half are less. After the

peer group feedback, this value rose to 8.25 (a

1.25-point increase).

Table 8: Comparing project Creativity results.

• Projects’ critical thinking outcome before

and after peer group feedback: As seen in

Table 9, for the initial method C1, 25% of

scores fall below the lower quartile value which

is 7. However, after peer group feedback, 25%

of scores fall below 8.25 (a 1.25-point

increase). For the second approach C2, 25% of

scores fall below the lower quartile value which

is 7. However, after peer group feedback, this

value rose to 9 (a 2-point increase).

Table 9: Comparing project Critical thinking results.

WEBIST 2023 - 19th International Conference on Web Information Systems and Technologies

160

6 DISCUSSION & CONCLUSION

The objective of this research was to examine the

impact of peer group feedback on assessment

outcomes within a PBCL context. Our research

strongly advocates for the implementation of peer

assessment within the PBCL context, leveraging

Ontology and the eLearning standards to optimize the

learning experience. Specifically, we relied on the

LOM eLearning standard to describe and annotate

learning resources, ensuring data reusability and

interoperability. Additionally, we employed the PAPI

LEARNER model to capture learners' performance,

personal information, portfolio details, preferences,

relations, and security aspects, creating an ontology-

driven model that underpinned our investigation. Our

approach involved two key components: the

formation of group peers and the implementation of

feedback mechanisms. To create peer, we utilized an

agglomerative clustering algorithm that considered

interactions between group members, aiming to

match groups with similar behaviors during the

project. This method was applied to half of our

learner groups, allowing us to compare its

effectiveness with a random pairing approach. The

second step involved facilitating real-time feedback

through chat rooms, where paired groups could

engage in providing and receiving feedback.

Comparing the two pairing methods, we discovered

that the agglomerative clustering strategy yielded

higher skill ratings compared to the random grouping

method. Specifically, the results indicated an 8.33%

increase in project validation proportions with the

agglomerative clustering strategy, compared to a

5.27% increase with the random strategy. In terms of

creativity, the random pairing approach resulted in a

1-point increase, whereas the second method yielded

a 1.25-point increase. Moreover, the second strategy

exhibited a 2-point improvement in critical thinking,

surpassing the 1.25-point improvement observed with

the random approach. These findings underscore the

significance of establishing intelligent pairings to

facilitate accurate and valuable feedback. The

behavior exhibited by group members emerges as a

crucial indicator of group dynamics. Future research

endeavors will explore additional criteria against

which our results can be benchmarked.

Regardless of the intelligent pairing strategy

employed, our findings consistently demonstrated the

immense benefits of peer group feedback for the

participating groups. The percentages of validated

final projects and the overall project scores exhibited

notable improvements. These enhancements

encompassed various assessed learning skills,

including critical thinking and creativity. The efficacy

of peer feedback as a formative method for enhancing

academic achievement aligns with the findings of

previous studies (Mcgrane & Hopfenbeck, 2020).

Furthermore, research conducted by authors (Saeedi

et al., 2021) has highlighted the positive impact of

peer feedback on student motivation, emphasizing its

value as an effective teaching technique. Group peer

feedback outperforms both no-group feedback and

sole instructor assessment. In line with contemporary

conceptions of formative assessment, our results

strongly advocate for the implementation of peer

assessment within a collaborative eLearning

environment, particularly when feedback is provided

synchronously. However, to further enhance the

quality of peer group feedback, our strategy can

benefit from integrating additional technologies. For

instance, we propose utilizing virtual reality to foster

increased interaction and conversation among group

pairings. Furthermore, automation of our peer group

feedback strategy is a goal we aspire to achieve,

incorporating it into the ongoing development of the

Intelligent Collaborative Assessment Framework

(ICAF).

In summary, our research underscores the

significance of peer group feedback in a PBCL

setting. It highlights the crucial role of intelligent

pairings and synchronous feedback in enhancing

learning outcomes. Moving forward, integrating

advanced technologies and automation will

contribute to further refinement and application of our

peer group feedback strategy, within the broader

framework of collaborative assessments.

REFERENCES

Bennett, R. E., & Service, E. T. (2014). Formative

assessment: A critical review. Assessment in

Education: Principles, Policy & Practice, April, doi

10.1080/0969594X.2010.513678.

Clark, I. (2012). Formative Assessment: Assessment Is for

Self-regulated Learning. Educational Psychology

Review, 24(2), 205–249. doi 10.1007/s10648-011-

9191-6.

Hwang, G., Zou, D., & Wu, Y. (2023). Learning by

storytelling and critiquing: A peer assessment-

enhanced digital storytelling approach to promoting

young students’ information literacy, self-efficacy, and

critical thinking awareness. Educational Technology

Research and Development, doi: 10.1007/s11423-022-

10184-y.

Topping, K. (1998). Peer assessment between students in

colleges and universities. Review of Educational

Research, 68(3), 249–276. doi 10.3102/00346543068

003249.

Ontology-Driven Intelligent Group Pairing in Project-Based Collaborative Learning

161

Dong, Z., Gao, Y., & Schunn, C. D. (2023). Assessing

students’ peer feedback literacy in writing: scale

development and validation. Assessment & Evaluation

in Higher Education, 1–16. doi

10.1080/02602938.2023.2175781.

Wu, Y., & Schunn, C. (2023). Passive, active, and

constructive engagement with peer feedback: A revised

model of learning from peer feedback. Contemporary

Educational Psychology, 73, 102160. doi:

10.1016/j.cedpsych.2023.102160.

Lin, G., & Lin, G. (2017). Anonymous versus identified

peer assessment via a Facebook-based learning

application: Effects on quality of peer feedback,

perceived learning, perceived fairness, and attitude

toward the system. Computers & Education, doi:

10.1016/j.compedu.2017.08.010.

Banister, C. (2020). Exploring peer feedback processes and

peer feedback meta-dialogues with learners of

academic and business English. Language Teaching

Research, 27(3), 746–764. doi 10.1177/136216882095

2222.

Cheniti Belcadhi, L. (2016). Personalized feedback for self-

assessment in lifelong learning environments based on

the semantic web. Computers in Human Behavior, 55,

562–570. doi 10.1016/j.chb.2015.07.042.

Chang, S. C., Hsu, T. C., & Jong, M. S. Y. (2020).

Integration of the peer assessment approach with a

virtual reality design system for learning earth science.

Computers & Education, 146, 103758. doi:

10.1016/j.compedu.2019.103758.

Kalmar, E., et al. (2022). The COVID-19 paradox of online

collaborative education: when you cannot physically

meet, you need more social interactions. Heliyon, 8(1),

e08823. doi: 10.1016/j.heliyon.2022.e08823.

Chorfi, A., Hedjazi, D., Aouag, S., & Boubiche, D. (2022).

Problem-based collaborative learning groupware to

improve computer programming skills. Behavior &

Information Technology, 41(1), 139–158. doi:

10.1080/0144929X.2020.1795263.

Mhlongo, S., Oyetade, K. E., & Zuva, T. (2020). The

Effectiveness of Collaboration Using the Hackathon to

Promote Computer Programming Skills. In 2020 2nd

International Multidisciplinary Information

Technology and Engineering Conference (IMITEC),

doi: 10.1109/IMITEC50163.2020.9334089.

Wang, X. M., & Hwang, G. J. (2017). A problem-posing-

based practicing strategy for facilitating students’

computer programming skills in the team-based

learning mode. Educational Technology Research and

Development, 65(6), 1655–1671. doi 10.1007/s11423-

017-9551-0.

Hadyaoui, A., & Cheniti-Belcadhi, L. (2023). An

Ontology-Based Collaborative Assessment Analytics

Framework to Predict Groups’ Disengagement. In

Intelligent Decision Technologies, 74–84.

Awuor, N. O., Weng, C., Piedad, E. J., & Militar, R. (2022).

Teamwork competency and satisfaction in online group

project-based engineering course: The cross-level

moderating effect of collective efficacy and flipped

instruction. Computers & Education, 176, 104357. doi

https://doi.org/10.1016/j.compedu.2021.104357.

Chen, C., & Du, X. (2022). Teaching and Learning Chinese

as a Foreign Language Through Intercultural Online

Collaborative Projects. Asia-Pacific Education

Research, 31(2), 123–135. doi 10.1007/s40299-020-

00543-9.

Aranzabal Maiztegi, A., Epelde Bejerano, E., & Artetxe

Uria, M. (2022). Team formation based on Belbin’s

roles to enhance students’ performance in project-based

learning. Education for Chemical Engineers, 38, 22–37.

doi: 10.1016/j.ece.2021.09.001.

Hadyaoui, A., & Cheniti-Belcadhi, L. (2022). Towards an

Ontology-based Recommender System for Assessment

in a Collaborative Elearning Environment. In WEBIST,

294–301. doi 10.5220/0011543500003318.

Ismail, F. A., Bungsu, J., & Shahrill, M. (2023). Improving

Students’ Participation and Performance in Building

Quantities through Think-Pair-Share Cooperative

Learning. Indonesian Journal of Educational Research

and Technology, 3(3), 203–216.

Murillo Zamorano, L. R., López Sánchez, J. Á., & Godoy

Caballero, A. L. (2019). How the flipped classroom

affects knowledge, skills, and engagement in higher

education: Effects on students’ satisfaction. Computers

& Education, 103608. doi: 10.1016/j.compedu.20

19.103608.

Senthil Kumaran, V., & Latha, R. (2023). Towards personal

learning environment by enhancing adaptive access to

digital library using ontology-supported collaborative

filtering. Library Hi Tech, ahead-of-print. doi:

10.1108/LHT-12-2021-0433.

García-González, H., L. Gayo, J. E., & Paule-Ruiz, M.

(2017). Enhancing e-learning content by using semantic

web technologies. IEEE Transactions on Learning

Technologies, 10(4), 544–550. doi 10.1109/TLT.20

16.2629475.

Hadyaoui, A., & Cheniti-Belcadhi, L. (2022). Towards a

context-aware personalized formative assessment in a

collaborative online environment. In Proceedings of

IEEE/ACS International Conference on Computer

Systems and Applications (AICCSA), 5–10. doi:

10.1109/AICCSA56895.2022.10017682.

Joy, J., Raj, N. S., & G., R. V. (2021). Ontology-Based E-

Learning Content Recommender System for

Addressing the Pure Cold-Start Problem. Journal of

Data and Information Quality, 13(3). doi 10.1145/

3429251.

Nouira, A., Cheniti-Belcadhi, L., & Braham, R. (2018). A

semantic web-based architecture for assessment

analytics. In Proceedings of the International

Conference on Tools with Artificial Intelligence

(ICTAI), 1190–1197. doi: 10.1109/ICTAI.2017.00181.

IEEE Standards Association. (2020). IEEE Standard for

Learning Object Metadata. IEEE Std 1484.12.1-2020,

1–50. doi: 10.1109/IEEESTD.2020.9262118.

Zine, O., Derouich, A., & Talbi, A. (2019). IMS Compliant

Ontological Learner Model for Adaptive E-Learning

Environments. International Journal of Emerging

WEBIST 2023 - 19th International Conference on Web Information Systems and Technologies

162

Technologies in Learning, 14(16). doi:

10.3991/ijet.14i16.10682.

Birgili, B. (2015). Creative and Critical Thinking Skills in

Problem-based Learning Environments. Journal of

Gifted Education and Creativity, 2(2), 71–80. doi:

10.18200/JGEDC.2015214253.

Jan, H., Noor-ul-Amin, S., & Matto, M. I. (2018). Modular

Object-Oriented Dynamic Learning Environment.

Journal of Applied Research in Education, 33(1), 1–11.

Ben-david, S. (2014). Understanding Machine Learning:

From Theory to Algorithms.

Ramos, L., Magaly, A., & Canuto, D. P. (2021). A

generalized average linkage criterion for Hierarchical

Agglomerative Clustering. Applied Soft Computing

Journal, 100, 106990. doi: 10.1016/j.asoc.2020.106

990.

McGrane, J. A., & Hopfenbeck, T. N. (2020). The Impact

of Peer Assessment on Academic Performance: A

Meta-analysis of Control Group Studies, 481–509.

Saeedi, M., Ghafouri, R., Tehrani, F. J., & Abedini, Z.

(2021). The effects of teaching methods on academic

motivation in nursing students: A systematic review, 1–

8. doi: 10.4103/jehp.jehp.

Ontology-Driven Intelligent Group Pairing in Project-Based Collaborative Learning

163