Managing Data Heterogeneity for Ontology-Driven Models:

Application to Gamified E-Learning Contexts

Yara Gomaa

1,2 a

, Christine Lahoud

, Marie-Hélène Abel

and Sherin Moussa

2,4 d

Laboratoire HEUDIASYC, Université de Technologie de Compiègne, Sorbonne Universités,

57 Avenue Landshut, Compiègne 60200, France

Laboratoire Interdisciplinaire de l’Université Française d’Egypte (UFEID Lab), Université Française d’Egypte,

21 Ismailia Desert Road Shorouk City, Cairo 11837, Egypt

CIAD UR 7533,Univ. BourgogneFranche-Comté, UTBM, F-90010 Belfort, France

Faculty of Computer and Information Sciences, Ain Shams University, Cairo, Egypt

Keywords: Gamification, E-Learning, Ontology, Data Collection, Data Integration.

Abstract: Data heterogeneity within gamified e-learning systems exposes a challenge for ontology-driven models,

specifically ontology-based recommender systems. These systems can help teachers who are unfamiliar with

gamification by offering personalized recommendations to gamify their pedagogical resources. Yet,

developing such systems requires collecting and integrating diverse data about users, resources, and game

elements, originating from multiple sources, like learning management systems and educational repositories,

each with varying formats and inconsistent semantics. This paper proposes an approach to manage the

complexities of collecting and preparing heterogeneous data for an ontology-driven model within gamified e-

learning contexts. A full overview is provided on the data workflow, which consists of two main phases: (1)

Data collection, which combines automated techniques through APIs and web scraping, and (2) Data

Integration by means of mapping the collected data into our Teacher in Gamified e-learning Context (TGC)

ontology to produce coherent and semantically enriched structure. The resulting data repository facilitates

semantic queries, inference, and knowledge enrichment, overcoming challenges like cold-start scenarios and

supporting the dynamic generation of personalized recommendations. This proposed approach aims to

establish a robust approach that addresses the challenges of data heterogeneity, ensuring consistent and

meaningful integration for ontology-based recommender systems in gamified e-learning contexts.

1 INTRODUCTION

Gamified E-learning systems have revolutionized

education by enhancing engagement and fostering

interactive learning experiences (Bennani et al., 2024;

Jumaa et al., 2017). Gamification, in this regard,

introduces game-like experiences to e-learning

environments through tools commonly known as

game elements . These game elements include points,

badges, leaderboards, etc (Maher et al., 2020).

Despite their potential, many teachers face challenges

in understanding and implementing gamification

https://orcid.org/0000-0001-9853-2416

https://orcid.org/0000-0002-4520-634X

https://orcid.org/0000-0003-1812-6763

https://orcid.org/0000-0001-9593-6909

effectively. This unfamiliarity hinders the ability to

optimize gamification to serve the purpose behind it

(Gomaa et al., 2024). Addressing this gap requires

systems that can assist teachers to link their objectives

with the relevant gamification strategy; by means of

recommending relevant game elements or gamified

resources. Ontology-driven models offer a promising

solution to this problem. By semantically structuring

data, such systems can provide a robust framework

for generating personalized suggestions tailored to

the pedagogical and gamification needs of teachers

(Bakhouyi et al., 2019). In addition, ontology-based

recommender systems can mitigate the cold-start

726

Gomaa, Y., Lahoud, C., Abel, M.-H. and Moussa, S.

Managing Data Heterogeneity for Ontology-Driven Models: Application to Gamiﬁed E-Learning Contexts.

DOI: 10.5220/0013496700003932

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Conference on Computer Supported Education (CSEDU 2025) - Volume 1, pages 726-736

ISBN: 978-989-758-746-7; ISSN: 2184-5026

problem by inferring recommendations through

established relationships, even with limited initial

data (Nymfodora-Maria Raftopoulou et al. 2023).

However, preparing data for such systems can be

challenging due to the heterogeneity of data. An

ontology-based recommender system expects

collecting data about three main categories, namely

(1) users, whether a teacher or learner, (2)

pedagogical resources, and (3) game elements. Each

category deals with further detailed data, for instance,

teachers are identified by personal information,

demographic data, institutional data, pedagogical and

gamification-related data. Such data can be found in

various sources, including educational repositories,

learning management systems, plugin libraries, and

user-generated sources, i.e., questionnaires (Joy et al.,

2021).

Preparing these diverse data sources for an

ontology-driven model demands a comprehensive

data collection and integration strategy:

• Data Collection: employing automated and semi-

automated techniques to gather the required data.

Automated methods rely heavily on APIs, which

provide structured access to data from platforms,

like Moodle

. These APIs allow for real-time or

scheduled retrieval of specific datasets. Another

technique is web scraping, which extracts the

HTML webpage content (Mansouri et al., 2022).

On the other hand, semi-automated methods often

include gathering qualitative data through

questionnaires. This method is helpful when

human input is needed, for example, for teachers

identifying their gamification experiences (Joy et

al., 2021).

• Data Integration: once collected, the

heterogeneous data must be aligned to a shared

ontology, which involves converting raw inputs

into Resource Description Framework (RDF)

triplets and mapping them onto the domain

concepts relevant to gamification and pedagogy.

Consistency checks and cleaning processes are

essential to reconcile conflicting data formats and

semantic inconsistencies (Villegas-Ch et al.

2023).

Numerous studies focused on data heterogeneity

in e-learning systems. However, limited focus was

given to addressing data heterogeneity in ontology-

driven models in gamified context. This leads to the

main research question of our study:

Moodle

How to collect and integrate heterogeneous data

from multiple sources for ontology-based models in

gamified e-learning contexts?

To address this question, we propose an approach

that systematically aggregates and integrates diverse

data into an ontology-driven model. Our

methodology focuses on collecting comprehensive

data from multiple sources, including educational

content, teacher profiles, and gamification elements.

By emphasizing the teacher’s role, we aim to make

them more familiar with applying gamification

techniques to their teaching practices. Additionally,

our approach allows for the dynamic adaptation of

game elements, meaning that educators can remove

or add specific game elements as needed to optimize

the learning experience. This flexibility ensures that

the full potential of gamification can be utilized,

enhancing both the educational process and the

learner’s engagement. The rest of the paper is

structured as follows: Section 2 explores the main

studies related to our work, involving data collection

and integration techniques. Followed by section 3 for

our proposed data collection and data integration

approach. Section 4 discusses the proposed work.

Then, section 5 presents the conclusion of our study,

highlighting the future directions to adopt.

2 RELATED WORK

This section explores the various data collection

techniques employed in the main studies within the e-

learning and gamified e-learning domains, focusing

on the types of data collected, the methods used for

their integration, and the role of ontologies in

structuring such data.

2.1 Data Management in E-Learning

Systems

This study focused on collecting and integrating

heterogeneous big data from E-Learning systems,

particularly Moodle LMS for analytics (Otoo-Arthur

& van Zyl, 2020). The integration process did not

involve ontology but relied on a big data framework

(biDeL) using Apache Spark, Hadoop and Flume for

data processing. The main goal was to enhance

learning analytics, enabling institutions to gain

insights into learner behavior and instructional

effectiveness. Data was collected through log

tracking, system interactions, and machine learning.

Challenges included data privacy, governance, and

Managing Data Heterogeneity for Ontology-Driven Models: Application to Gamiﬁed E-Learning Contexts

727

the complexity of handling high volume, high

velocity and varied data sources, necessitating

advanced frameworks for scalability and security.

For enhancing the learning analytics within an e-

learning systems such as MOOCs, data collected was

mainly about the learners’ interactions, profiles, and

engagement within the e-learning systems. These data

were collected from Experience API (xAPI) and

IEEE Standard for Learning Technology (SLT) (Del

Blanco et al., 2013).

The previous studies focused on the non-gamified

context, limiting learner preferences, datatypes

collected, and enhanced learning outcomes. Unlike

these studies, where collected data was primarily used

for learning and analytics, our approach focuses on

structuring data within an ontology, enabling

continuous refinement, semantic enrichment, and

enhanced knowledge representation.

2.2 Data Management in

Ontology-Based E-Learning

Systems

This paper presents a semantic web-based approach

to enhance E-Learning systems interoperability using

RDF and next generation SCORM specifications

(Bakhouyi et al., 2019). It addresses interoperability

challenges in LMS platforms like Moodle by

transforming JSON data into RDF with JSON-LD,

ensuring seamless data sharing between xAPI and

LMS. The study highlights Semantic Web

Technologies’ potential to improve content exchange,

tracking, and analytics, enabling better accessibility,

learning personalization, and integration of mobile

applications into e-Learning ecosystem.

The study proposes a hybrid Adaptive

Educational e-Learning System (AEeLS) integrating

artificial intelligence and semantic web technologies

to personalize learning (Demertzi & Demertzis,

2020). AEeLS adapts educational content based on

student skills and experience by using ontology

matching and a recommendation system. The

ontology matching employs semi-supervised

machine learning to align educational resources,

while the recommendation mechanism uses

collaborative and content-based filtering to suggest

relevant learning materials. The system improves

interoperability, efficiency, and adaptability in e-

learning. Experiments on datasets from OAEI 2014,

ADRIADNE, and MERLOT demonstrate their super

superior performance in matching and recommending

educational content.

Resources were considered in more detail in

(Bouihi & Bahaj, 2019), focusing on data collection

and integration related to learning content, learner

profiles, social interactions, and learner activities to

enhance personalized learning through a semantic

web-based recommendation system. It used learning

objects modelled with standards like LOM and

SCORM as the core data, enriched with contextual

information like learner’s history, performance, and

social connections, and represented using ontologies

like FOAF. Accordingly, these data were utilized to

personalize recommendations for learners. However,

the study could have benefited from extending FOAF

ontology to further model the various parameters of a

resource and a learner.

The challenges of data interoperability in

collaborative e-learning systems were tackled in

(Masud, 2016), focusing on how to manage and share

learning content across different systems with

varying schemas. The data collected included

learning content and its metadata and shared between

independent e-learning systems using schema-level

and data-level mappings. Through query processing,

relevant resources were retrieved based on user

queries. This study enforced collaboration and

sharing of resources between learners and teachers, as

the resources were semantically categorized and

managed, ensuring interoperability.

However, teachers-related data was explored in a

limited context. This was overcome in (Nashed et al.,

2022), in which an ontology-based approach was

introduced to integrate diverse educational data

related to teacher’s personal, organizational, and

contextual environments. It collected data about

educational resources and mapped to Teacher-

Context Ontology (TCO) using D2RQ mapping,

which offered a comprehensive view of the teacher’s

professional and personal context. The approach used

automated mapping techniques to merge data from

various sources, facilitating its use in personalized

educational systems, like recommender systems.

Nevertheless, gamification in the e-learning context

was not handled, where recommendations were fixed

to the e-learning aspect only.

2.3 Data Management Gamified

Context

Data heterogeneity in gamified e-learning systems

was the focus of various studies. A gamified

educational system was presented in (Nymfodora-

Maria Raftopoulou et al. 2023) to enhance the

gamified learning experience. It collected data, such

as user experience rating, exercise completion times,

success rates, and correct/incorrect answers, and used

machine learning algorithms to analyze data, identify

EKM 2025 - 8th Special Session on Educational Knowledge Management

728

patterns in learner behaviour, and provide

personalized feedback. The system used collaborative

filtering and content-based filtering to recommend

future exercises that meet each learner’s strengths and

weaknesses.

The authors focused on collecting and integrating

student interaction data within a gamified e-learning

system designed to develop computational thinking

(Villalba-Condori et al., 2022). The system employed

supervised learning techniques, specifically Support

Vector Machines (SVM), to classify learning

progress and recommend personalized content. The

data came from students’ in-game activities tracking

performance, engagement, and decision making.

While the study did not explicitly integrate an

ontology-based approach, it leveraged machine

learning and statistical models for recommendations

and performance analytics. Challenges the

complexity of accurately modeling student learning

behaviors in a gamified environment.

Another adaptive gamification recommendation

system was introduced in (Bennani et al., 2024),

focusing on game elements adaptation to improve

interactive learning environments. It collected both

static and dynamic data about learners, such as their

profile information, interaction, and performance

metrics. Player type tests were used to assess the

learner’s preferred game elements. The integration

process involves a matrix factorization process, in

which the data is then utilized in collaborative

filtering recommender system.

In (Alonso-Fernández et al., 2020), data was

collected to predict learner’s knowledge outcomes

based on their interactions within a serious game.

Data included personal information, like age and

gender, knowledge data through pre and post-tests,

and game interaction data, which were tracked

through Experience API, such as scores, levels

completed, etc.

Considering the presented studies, despite the

progress in data integration within e-learning

systems, a critical gap remains in addressing the

heterogeneity of data across game elements, teachers,

and the broader e-learning context. Most of the

presented studies have primarily focused on learners

and educational resources, with minor attention given

to the teacher’s profile or the varying data related to

game elements. In particular, the complexities

introduced by the teacher’s context and gamification

aspects have not been thoroughly investigated. This

oversight highlights a significant opportunity to

develop comprehensive systems that adopt teacher-

specific data and game elements, enabling a holistic

approach to a personalized and gamified e-learning

experience.

3 PROPOSED APPROACHES

To prepare data for an ontology-driven model, it

undergoes two main processes: (1) data collection and

(2) data integration via ontology mapping. In this

section, we introduce our proposed approach for the

data collection process, the data sources, and the

formats used to gather and store data.

3.1 Data Management Process

As the main goal of this study is to harmonize

heterogeneous data within an ontology-driven model

in gamified e-learning contexts, this section

highlights the data workflow within a recommender

system, along with its categories and data sources.

Figure 1 provides a high-level overview of the entire

data management process, including data collection,

integration and their utilization within ontology-

driven models such as ontology-based recommender

system in gamified e-learning contexts. The process

begins with collecting data about users (teachers and

learners), resources, and game elements. For the

users, data is broadly categorized into personal

information, pedagogical and gamification-related

data, and activity log. Users’ personal information

includes their name, age, gender, registered

educational institution, and preferred language.

Regarding the teacher’s pedagogical-related

information, data is related to the teaching style and

pedagogical experience, to determine how skillful the

teacher is in applying teaching strategies. In addition

to the learning objectives, which define the teacher’s

intention behind a particular resource, in terms of the

expected learning outcomes. As for the gamification-

related data for teachers, it includes gamification

experience, which reflects their ability to effectively

utilize gamification techniques. The teacher’s

behavioral objectives refer to the targeted behavior

expected from the learner towards a presented

resource and playing style. On the other hand, the

learner’s pedagogical-related data refers to the

learner’s learning style, reflecting the learner’s

preferred way to receive and process the presented

information. The learner’s gamification-related data

includes their player type, which helps identify the

most effective methods to maintain their engagement.

In addition, the activity log refers to the teacher’s and

learner’s interactions throughout the system,

including their feedback, ratings, and behavior.

Managing Data Heterogeneity for Ontology-Driven Models: Application to Gamiﬁed E-Learning Contexts

729

Figure 1: Data Management for Ontology-driven Model.

Concerning resources, the required data mainly

includes the related educational topic, the resource’s

type (video, word document, audio file, etc.), in

addition to the game elements associated with that

resource (if any). The available game elements need

to be identified with basic information, like title,

description of its functionality, along with popularity

metrics, like total downloads and likes, etc. The game

element’s version indicates whether it is up-to-date or

obsolete, and finally, a reference to the game element,

such as URL, can be provided as well. After gathering

the needed data, the next step is to pre-process and

clean data to ensure consistency with other data

sources. This involves standardizing the format of the

collected data and handling any missing or

incomplete information. Once cleaned, data is

mapped to the ontology to ensure it could be

effectively queried and utilized in a recommendation

process. Having the data integrated into our

ontological model, it is utilized by recommender

systems to provide personalized recommendations

for teachers. For the following sections, we adopt a

unified scenario as an example of how the data are

collected and integrated for recommender systems.

This scenario is illustrated in two partitions as

scenario-part 1 and scenario-part 2 in sections 3.1 and

3.2 respectively.

Scenario Overview: Mr. Robert Sedgewick is a

teacher who is unfamiliar with gamification but

wishes to integrate game elements into his course on

“Computer Science Programming”. To demonstrate

the data collection and integration processes, an

existing Coursera course entitled as “Computer

Science: Programming with a Purpose” is adopted as

a case study. This course has predefined data, such

as the course modules and resources. Moreover, for

the gamification strategies, game element plugins are

adopted from Moodle. This demonstrates how the

relevant data are collected and integrated to be ready

for utilization by ontology-based recommender

systems.

3.2 Data Collection

The data collection process involves extracting and

aggregating data from several key categories.

Referring to our scenario to demonstrate the data

collection process:

Scenario-part1: “To provide Mr. Robert Sedgewick

with personalized recommendations on the

gamification of his course, his personal data,

pedagogical-related data, gamification-related data,

and activity log are collected. In addition, the

targeted audience knowledge level, and the learning

objectives are collected, as well as the course’s

structure itself along with its modules and resources

from Coursera.”.

Table 1 presents the typical data sources for each

category, along with their expected data formats.

Educational institutions include universities, schools,

etc., whereas educational platforms are e-learning

management systems, such as Moodle. Users refer

mainly to teachers and learners, where various types

of data are required from them, scattered across

various data sources. Typical personal information

can be collected from their educational institutions or

from their user profile on the educational platforms.

However, there are other data related to pedagogy and

gamification that can be collected through

questionnaires, as they are particularly difficult to

find explicitly through educational platforms or

institutions. Such data includes teacher’s teaching

EKM 2025 - 8th Special Session on Educational Knowledge Management

730

style, teacher’s playing style, learner’s learning style,

and learner’s player type. These data can be initially

identified through questionnaires like the Fedler

Silverman Learning Style Model FSLM (Felder &

Spurlin, 2005), or Marczewski’s player type model

(Tondello et al., 2016). Furthermore, activity logs

refer to the recorded user’s interaction activity,

determining the user’s behavior, feedback, accessed

resources, ratings, etc. However, the user’s

interactivity within an educational platform can only

be accessed and collected through that platform.

Furthermore, Resources are shared through digital

form, which can be accessible through educational

platforms or institutions.

Table 1 Data types and their corresponding sources for

ontology integration.

Data Type Data Source

Data

Format

User

Personal

Information

Educational

Institutions,

Educational

Platforms

CSV,

JSON

Pedagogy-

related Data

Online

questionnaire

CSV,

JSON,

database

Gamification-

related Data

Online

questionnaire

CSV,

JSON,

database

Activity Log

Educational

Platforms

CSV,

JSON,

TXT,

XML

Resources Educational

Institutions,

Educational

Platforms

Educational

Institutions,

Educational

Platforms

CSV,

JSON

Game

Elements

Educational

Platforms

Educational

Platforms

XML

Data are gathered about a programming course on

Coursera

that was done semi-automatically through

web scrapping by BeautifulSoup API. As shown in

the resulted JSON file in Figure 2, it contains basic

information about the course, including course title,

course instructors, the course language, the targeted

level of audience, skills associated with the course,

and the course modules. Each module has title and a

set of resources, where each resource is identified by

its type and duration. On the other hand, game

elements are expected to be associated to a resource.

Game elements are plugins that can be shared across

educational platforms. In this regard, Moodle

gamification plugins are used as a case study for data

Coursera

collection of game elements. Figure 3. shows a

screenshot from Moodle demonstrating part of the list

of game elements in the platform, with detailed

description for each one.

Figure 2: Part of JSON file for the course, teachers, lessons,

and resources through BeautifulSoup.

Figure 3: Screenshot from Moodle for game elements

plugins.

This data were then collected in JSON file as shown

in Figure 4, making it ready to be integrated into our

ontology.

3.3 Data Integration

To harmonize the data gathered from the data

collection phase, data is mapped to our Teachers in

Gamified e-learning Context (TGC) Ontology. The

following scenario demonstrates the work done in the

data integration process:

Scenario-part 2: “The collected data is then

integrated to our TGC ontology, where the structured

Managing Data Heterogeneity for Ontology-Driven Models: Application to Gamiﬁed E-Learning Contexts

731

data of Mr. Robert, the game elements from Moodle,

and the course are mapped to our TGC ontology

based on the defined rules. The resultant ontology-

integrated data is in RDF format.”

TGC ontology focuses on the teacher’s

perspective within the gamified e-learning context. It

reuses the existing ontologies MC2 and TCO (Abel et

al., 2007; Nashed et al., 2021). MC2 ontology is

concerned with the collaboration and shareability

between resources, while TCO is concerned with

defining the teacher in several contexts, including

working and living environments. Our proposed TGC

ontology utilizes MC2 and TCO to have a

comprehensive semantic structure for the

gamification aspect, together with the teachers and

pedagogical resources. The automated technique used

for mapping the collected data to our TGC ontology

is carried out through RDFLib, a python library for

working with RDF data. The mapping is performed

by defining specific rules as shown in (1) and (2):

Raw Data  Instructors{Robert Sedgewick,

evin Wa

}

(1)

Raw Data  Course Title: “Computer Science:

Pro

rammin

with a Pur

ose”

(2)

Figure 5 shows that all attributes with the name

Instructors are mapped to TCO:Teacher, where the

values of Instructors are of type xsd:String, mapped

to the data property TCO:Teacher.has_name.

Similarly, Course Title is mapped to

TCO:Course.has_title, where the object property

TCO:teaches associates the TCO:Teacher with

TCO:Course, as shown in (3):

RDF Triplet  Robert Sedgewick, Teaches,

Computer Science: Programming with a

Pur

ose

(3)

Figure 4: Part of JSON file for the game elements through BeautifulSoup.

Figure 5: Teachers mapped automatically to RDF through RDFLib.

EKM 2025 - 8th Special Session on Educational Knowledge Management

732

Moving to the course’s modules and resources,

Figure 6 shows the module mapped to TCO:Lesson,

whereas its activities mapped to “TCO:Resource”,

associated together through the object property

“TCO:contain_Resource”. On the other hand, game

elements collected were mapped to our TGC

ontology. To understand the ontological structure of

a game element, Figure 7. shows the game element

class with its associated data properties.

“TGC:GameElementResource” class has several

performance metrics, like total downloads, total likes,

total sites using this plugin, the plugin’s latest release,

and the latest version. These data are particularly

helpful for a recommender system to recommend an

appropriate game element that is relevant, popular,

and up to date. The mapping of the game elements is

performed by defining rules that associate each value

attribute from the JSON file to a

TGC:GameElementResource as shown in (4) and (5):

Raw Data  title: “Block Game” (4)

Raw Data  published: "Wed,31 May 2023

12:55:01 GMT"

(5)

All attributes with the name title are considered as

xsd:String and are mapped to the data property

TGC:GameElementResource.hasTitle. For the

attribute named published, it displays the date of

publishing the plugin, however, our TGC ontology

requires the duration in months since the latest

published date. Therefore, it is pre-processed before

the ontology mapping process. Figure 8 shows the

resulted RDF file, where it illustrates that the game

element has title “stash” and it is a named individual

of type “TGC:GameElementResource” with all

fetched data properties includes the resource title, its

release date, a short description about this resource,

and the resource’s plugin Uniform Resource

Identifier (URI) as shown in (6):

RDF Triplet  Block_Game,

hasLatestRelease, "20"^^xsd:nonNegativeIntege

(6)

Figure 6: Resources mapped automatically to RDF through RDFLib.

Figure 7: GameElementResource ontology class with its data properties.

Managing Data Heterogeneity for Ontology-Driven Models: Application to Gamiﬁed E-Learning Contexts

733

Figure 8: Game Elements mapped automatically to RDF through RDFLib.

4 DISCUSSION

In this research two main data sources are utilized to

formalize the teacher's perspective within the given e-

learning contexts.

The data related to the teacher and educational

content including core structure lessons and resources

were sourced from Coursera. Meanwhile, the game

elements were treated as independent standalone

plugins adopted from Moodle. This data was

successfully structured and integrated within our

proposed TGC ontology. Providing them with

semantic meaning and associating them under a

coherent unified structure.

Previous studies have aimed to structure data

primarily. Focusing on learners with some studies

incorporating aspects of course structure, and others

focusing more on resource details. Other studies have

explored game elements in combination with learner

data leaving out the teacher's context and course data.

This created a gap in linking these various elements.

This paper takes an initial step toward bridging this

gap by combining detailed data on users (teachers and

learners), courses, and game elements into a single

unified structure.

This study has encountered several challenges,

firstly the limited scope of the game elements

presents a constraint. These elements are typically

integrated within e-learning systems. Standalone

game elements are scarce, with limited variety and

accessible through limited data sources. This narrow

range may potentially limit the richness and

adaptability of recommendations. In addition to this,

Moodle platform introduced another challenge, as

retrieving data automatically proved difficult. We

attempted Moodle APIs, but this approach was not

successful as the access was not authorized. The only

accessible data was through partial web scraping,

since not all properties of the game elements could be

retrieved. Moreover, key teacher-related properties

such as player type, teaching style, and gamification

experience, are not typically included in teacher

profiles in educational platforms or institutions. As a

result, this data could not be retrieved automatically.

Instead, questionnaires were necessary to obtain

initial insights into teachers’ profiles. However, these

questions require thorough revision to ensure the

validity of the questions which adds to the workload

and necessitates the involvement of domain experts.

Therefore, this approach would benefit from

further refinement. One key area for improvement is

that the data were single sourced in this study, which

avoided issues related to that data fusion process. In

real-word applications, these data often come from

diverse platforms requiring careful integration.

Additionally, some teacher related data must be

collected through validated questionnaires to ensure

accuracy and comprehensiveness and aspect that

should be addressed in future improvements.

5 CONCLUSIONS

Data heterogeneity within gamified e-learning

systems is quite challenging for ontology-based

recommender systems. Teachers, learners, resources,

and game elements typically originate from multiple

sources, ranging from learning management systems

and educational repositories to questionnaires and

APIs, where each has varying formats and

inconsistent semantics. Thus, driving our main

research question for this study, that is how to collect

and integrate heterogenous data from multiple

sources for ontology−based models in gamified

e−learning contexts. For that, an approach was

proposed to manage the complexities of collecting

EKM 2025 - 8th Special Session on Educational Knowledge Management

734

and preparing heterogeneous data in ontology-based

recommender systems for the gamified e-learning

contexts. It proposes automated techniques to collect

data from various sources, like Moodle and Coursera

and maps them to our proposed Teacher in Gamified

e-learning Context (TGC) ontology. Future research

directions include development and refinement of

semi-automated data collection methods, such as

structured questionnaires, to gather comprehensive

information about teachers’ playing style, teaching

style, and pedagogical and gamification experience.

These questionnaires could be validated through pilot

testing with representative teacher samples to ensure

content validity and reliability. Data collection in this

regard will be validated through a semi-automated

process to ensure completeness and conciseness.

Particular attention should be given to ensuring the

content validity and reliability of these

questionnaires, as well as the systematic validation of

the collected data to enhance the accuracy and

relevance of recommendations provided by ontology-

based recommender systems. Thus, the development

of such ontology-based recommender systems

utilizing these integrated data is essential to provide

personalized recommendations that align with

teachers’ needs and enhance the gamification process

in e-learning environments. This involves leveraging

the ontology to infer meaningful relationships

between teacher profiles, resources, and game

elements.

REFERENCES

Abel, E., Benayache, A., Lenne, D., & Moulin, C.

(2007). E-MEMORAe: A Content-Oriented

Environment for E-Learning. http://www.altrc.org/

specification.asp.

Alonso-Fernández, C., Martínez-Ortiz, I., Caballero,

R., Freire, M., & Fernández-Manjón, B. (2020).

Predicting students’ knowledge after playing a

serious game based on learning analytics data: A

case study. Journal of Computer Assisted Learning,

36(3), 350–358. https://doi.org/10.1111/jcal.12405

Bakhouyi, A., Dehbi, R., Banane, M., & Talea, M.

(2019). A semantic web solution for enhancing the

interoperability of E-learning systems by using next

generation of SCORM specifications. International

Journal of Emerging Technologies in Learning,

14(11), 174–185. https://doi.org/10.3991/

IJET.V14I11.10342

Bennani, S., Maalel, A., Ghezala, H. Ben, & Daouahi,

A. (2024). Towards an Adaptive Gamification

Recommendation Approach for Interactive

Learning Environments. In Lecture Notes on Data

Engineering and Communications Technologies

(Vol. 199, pp. 341–352). Springer Science and

Business Media Deutschland GmbH.

https://doi.org/ 10.1007/978-3-031-57840-3_31

Bouihi, B., & Bahaj, M. (2019). Ontology and rule-

based recommender system for e-learning

applications. International Journal of Emerging

Technologies in Learning, 14(15), 4–13.

https://doi.org/10.3991/ ijet.v14i15.10566

Del Blanco, A., Serrano, A., Freire, M., Martinez-Ortiz,

I., & Fernandez-Manjon, B. (2013). E-Learning

standards and learning analytics. Can data

collection be improved by using standard data

models? IEEE Global Engineering Education

Conference, EDUCON, 1255–1261.

https://doi.org/10.1109/ EduCon.2013.6530268

Demertzi, V., & Demertzis, K. (2020). A Hybrid

Adaptive Educational eLearning Project based on

Ontologies Matching and Recommendation System.

Felder, R. M., & Spurlin, J. (2005). Applications,

Reliability and Validity of the Index of Learning

Styles*.

Gomaa, Y., Moussa, S., Lahoud, C., & Abel, M. H.

(2024). Exploring Recommender Systems for

Assisting Teachers in E-Learning Gamification.

Procedia Computer Science, 246(C), 2312–2321.

https://doi.org/ 10.1016/j.procs.2024.09.556

Joy, J., Raj, N. S., & Renumol, V. G. (2021). Ontology-

based E-learning Content Recommender System

for Addressing the Pure Cold-start Problem.

Journal of Data and Information Quality, 13(3).

https://doi.org/ 10.1145/3429251

Jumaa, Y. M., Moussa, S. M., & Khalifa, M. E. (2017).

The Main Aspects of Adaptive Educational Games

for Normal and Disabled / Disordered Learners : A

Comprehensive Study. Icicis, 348–355.

Maher, Y., Moussa, S. M., & Khalifa, M. E. (2020).

Learners on Focus: Visualizing analytics through

an integrated model for learning analytics in

adaptive gamified e-learning. IEEE Access,

197597–197616. https://doi.org/10.1109/

ACCESS.2020.3034284

Mansouri, K., Kaiss, W., & Poirier, F. (2022).

Personalized e-learning recommender system based

on a hybrid approach. IEEE Region, 8.

https://doi.org/10.1109/

EDUCON52537.2022.9766650ï

Masud, M. (2016). Collaborative e-learning systems

using semantic data interoperability. Computers in

Human Behavior, 61, 127–135.

https://doi.org/10.1016/ j.chb.2016.02.094

Nashed, N. N., Lahoud, C., & Abel, M. H. (2021).

TCO : a Teacher Context Ontology. Proceedings of

the 2021 IEEE 24th International Conference on

Computer Supported Cooperative Work in Design,

CSCWD 2021, 757–762.

https://doi.org/10.1109/CSCWD49262.2021.

9437622

Nashed, N. N., Lahoud, C., Abel, M.-H., & ene Abel,

M. (2022). Ontology-based teacher-context data

Managing Data Heterogeneity for Ontology-Driven Models: Application to Gamiﬁed E-Learning Contexts

735

integra-tion. https://hal.archives-ouvertes.fr/hal-

03537280

Nymfodora-Maria Raftopoulou, & Petros L. Pallis.

(2023). Gamified Learning Systems’ Personalized

Feedback Report Dashboards via Custom Machine

Learning Algorithms and Recommendation

Systems. Sociology Study, 13(3). https://doi.org/

10.17265/2159-5526/2023.03.006

Otoo-Arthur, D., & van Zyl, L. , T. (2020). A Scalable

Heterogeneous Big Data Framework for e-

Learning Systems. IEEE.

Tondello, G. F., Wehbe, R. R., Diamond, L., Busch, M.,

Marczewski, A., & Nacke, L. E. (2016). The

gamification user types Hexad scale. CHI PLAY

2016 - Proceedings of the 2016 Annual Symposium

on Computer-Human Interaction in Play, 229–243.

https://doi.org/10.1145/2967934.2968082

Villalba-Condori, K., Vera-Sancho, J., Vera-Vasquez,

C., Mamani-Calcina, J., & Cardona-Reyes, H.

(2022). Knowledge-Based Recommender System in

a Gamified Environment for Computational

Thinking Development.

EKM 2025 - 8th Special Session on Educational Knowledge Management

736