A VARK Learning Style Based Recommendation System for Adaptive E-learning
No Thumbnail Available
Date
2026
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Saudi Digital Library
Abstract
In numerous conventional platforms of e-learning, there has been constant delivery of identical learning content to all learners, regardless of individual differences in terms of learning style, prior knowledge, or level of understanding. The above one-size-fits-all approach can limit learner engagement and reduce the effectiveness of online learning environments. This emphasises the need for adaptive e-learning in offering more effective recommendations of learning resources based on the requirements of the learners, and by considering many factors that affect learning, such as student learning style, knowledge level, personality and the time available etc. This thesis presents a novel system of adaptive e-learning, which incorporates the VARK model of learning styles with a mechanism of recommendations to help deliver tailored materials of learning. Specifically, the system identified the learner’s style using Visual, Auditory, Reading/Writing, and Kinesthetic (VARK) assessment before conducting a pre-test to detect gaps in knowledge. As a result, students received ranked content for personalised learning, followed by a post-test to measure improvement. Two studies (Pilot and Main) were conducted at King Khalid University in Saudi Arabia to assess the efficiency of the system. The pilot study appraised the test methodology and the system usability and functionality with two groups of students. The main study also involved two main groups, namely, an experimental group using the adaptive system and a control group using a non-adaptive e-learning platform. Pre- and post-test scores, along with engagement and feedback measures, were employed to determine the impact of the system and demonstrated measurable gains. Thus, the main contributions of the current investigation included the design and development of a VARK learning style based of recommendation system and the interpretation of a methodology for evaluating personalised learning. The results revealed that the above approach could help enhance students’ comprehension, satisfaction, engagement, and motivation compared to standard methods of e-learning.
Description
This thesis investigates the design, development, and evaluation of an adaptive e-learning recommendation system that personalises learning content based on students' VARK learning styles and knowledge levels. The system, named Flex-Learning, integrates three key components: a VARK learning style assessment to identify each student's preferred mode of receiving information (visual, auditory, read/write, or kinaesthetic), a pre-test to assess knowledge gaps across course topics, and a fuzzy logic engine combined with rule-based classification to generate a ranked list of personalised learning materials that addresses both the format and priority of content delivery. The system was built using React.js, Node.js, and a JSON-based database, following a Waterfall software development methodology.
The research was motivated by the limitations of conventional e-learning platforms, which deliver identical content to all learners without accounting for individual differences in how students learn or what they already know. This challenge was particularly evident in Saudi Arabian higher education, where COVID-19 disruptions, lecturer shortages, and the national Vision 2030 digital transformation agenda highlighted the need for more effective online learning solutions.
The system was evaluated through two studies conducted at King Khalid University in Saudi Arabia. A pilot study with 31 students validated the usability and functionality of the system, leading to refinements before the main study. The main study involved 82 students (42 experimental, 40 control) and 7 teachers, employing a randomised controlled experimental design with a mixed-method approach combining pre/post-tests, Likert-scale questionnaires, and qualitative feedback. Non-parametric statistical tests (Mann-Whitney U and Wilcoxon Signed-Rank) were used to test seven hypotheses, of which five were fully confirmed and two partially confirmed.
The results demonstrated that students using the adaptive system achieved significantly higher post-test scores than those using the non-adaptive version (p < .001), with high levels of satisfaction across all measured factors and positive endorsement from teachers. The research contributes a novel integration of VARK, fuzzy logic, and rule-based classification in a single adaptive system, evaluated in a real educational setting within Saudi Arabian higher education.
Keywords
Adaptive e-learning, recommendation system, VARK learning style, learning impact.