PREDICT LEARNERS’ PERFORMANCE USING AN ONTOLOGICAL-BASED MODEL ON AN E-LEARNING PLATFORM

In learning analytics and educational data mining, a prominent challenge is posed by the lack of portability and transferability of predictive models across different courses. A novel ontology-based decision tree model is introduced in this study, which significantly enhances portability by incorpor...

Full description

Saved in:
Bibliographic Details
Main Author: Safa Ridha Albo Abdullah
Format: Article
Language:English
Published: Faculty of Engineering, University of Kufa 2025-07-01
Series:Mağallaẗ Al-kūfaẗ Al-handasiyyaẗ
Subjects:
Online Access:https://journal.uokufa.edu.iq/index.php/kje/article/view/17337
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In learning analytics and educational data mining, a prominent challenge is posed by the lack of portability and transferability of predictive models across different courses. A novel ontology-based decision tree model is introduced in this study, which significantly enhances portability by incorporating semantic features. Unlike conventional decision tree models that are static and course-specific, the approach allows for the dynamic generation and adaptation of decision tree rules using ontologies, thus enabling seamless application across multiple courses without compromising accuracy. The core innovation of the method lies in the use of the Semantic Web Rule Language (SWRL) to integrate decision tree rules into an ontological framework, allowing reasoning and adaptation based on domain-specific semantics. By embedding decision tree rules that differentiate between pass and fail outcomes within an ontology, superior portability, and adaptability are achieved across courses with similar usage patterns. Applied to the Open University Learning Design (OULD) science courses dataset, advanced machine learning techniques were utilized to derive predictive rules, achieving an exceptional prediction accuracy of 97%. This demonstrates not only the effectiveness of the model but also its potential for robust transferability across various educational contexts, marking a significant advancement over existing methods
ISSN:2071-5528
2523-0018