Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Aug 20:14:809.
doi: 10.12688/f1000research.166248.1. eCollection 2025.

Personalised and Collaborative Learning Experience (PCLE) Framework for AI-driven Learning Management System (LMS)

Claireta Tang Weiling  1 , Lew Sook Ling  2 , Ooi Shih Yin  2
Affiliations

Personalised and Collaborative Learning Experience (PCLE) Framework for AI-driven Learning Management System (LMS)

Claireta Tang Weiling et al. F1000Res. .

Abstract

Background: Understanding student engagement and academic performance is crucial in AI-driven e-learning environments. Many learning management systems (LMS) lack effective collaborative course recommendation strategies, limiting support for personalised learning experiences.

Methods: This study developed and evaluated collaborative filtering and machine learning models to generate course recommendations. Machine learning models such as K-Nearest Neighbours (KNN), Singular Value Decomposition (SVD), and Neural Collaborative Filtering (NCF) were applied. Two education-related datasets from Kaggle were used. The first contains 100,000 course reviews from Coursera, and the second dataset includes 209,000 course details and comments from Udemy. Data preprocessing was conducted to clean and structure both datasets. The model effectiveness was evaluated using Mean Absolute Error (MAE), Hit Rate (HR), and Average Reciprocal Hit Ranking (ARHR).

Results: K-Nearest Neighbours showed the highest performance on the Coursera dataset, while Singular Value Decomposition and Neural Collaborative Filtering maintained stable predictive accuracy across both datasets. The findings indicate that dataset characteristics influenced model performance. K-Nearest Neighbours worked effectively with structured and consistent data, while Singular Value Decomposition and Neural Collaborative Filtering produced consistent outcomes across diverse datasets.

Conclusions: This study contributes to e-learning research by demonstrating the potential of collaborative filtering and machine learning in enhancing course recommendations and promoting engagement in the learning management system. Limitations include the use of two datasets and a limited set of machine learning models. Future work aims to integrate learning styles and evaluate the framework across more diverse educational contexts to support adaptive and collaborative learning.

Keywords: Collaborative Filtering; K-Nearest Neighbours Model (KNN); Learning Management System; Neural Collaborative Filtering Model (NCF); Personalised Learning; Singular Value Decomposition Model (SVD).

PubMed Disclaimer

Conflict of interest statement

No competing interests were disclosed.

Figures

Figure 1.
Figure 1.. Felder and Silverman Learning Style model ( Rasheed & Wahid, 2021).
An overview of the dimensions of learning preferences as defined in the Felder-Silverman model.
Figure 2.
Figure 2.. Learning management system ( Khalaf et al., 2022).
A conceptual representation of the components and workflow within a learning management system (LMS).
Figure 3.
Figure 3.. Kolb’s Learning Style model ( Sanjabi & Ali Montazer, 2020).
The four-stage learning cycle and associated learning styles proposed by Kolb.
Figure 4.
Figure 4.. Proposed framework.
The framework integrates collaborative filtering and machine learning to enhance e-learning outcomes.
Figure 5.
Figure 5.. Check missing values for dataset 1.
Visualisation showing missing data distribution in the Coursera dataset.
Figure 6.
Figure 6.. Check missing values for dataset 2.
Visualisation showing missing data distribution in the Udemy dataset.
Figure 7.
Figure 7.. Remove missing values for dataset 2.
Final version of the Udemy dataset after data cleaning.
Figure 8.
Figure 8.. Recommended course in KNN model for dataset 1.
Top course recommendations generated by the K-Nearest Neighbours model based on Coursera data.
Figure 9.
Figure 9.. Recommended course in KNN model for dataset 2.
Top course recommendations generated by the K-Nearest Neighbours model based on Udemy data.
Figure 10.
Figure 10.. Recommended course in SVD model for dataset 1.
Top course recommendations generated by the Singular Value Decomposition model based on Coursera data.
Figure 11.
Figure 11.. Recommended course in SVD model for dataset 2.
Top course recommendations generated by the Singular Value Decomposition model based on Udemy data.
Figure 12.
Figure 12.. Recommended course in NCF model for dataset 1.
Top course recommendations generated by the Neural Collaborative Filtering model for Coursera data.
Figure 13.
Figure 13.. Recommended course in NCF model for dataset 2.
Top course recommendations generated by the Neural Collaborative Filtering model for Udemy data.
See this image and copyright information in PMC

References

    1. Adona JCM: 100K Coursera’s Course Reviews Dataset. 2018. Accessed 10 12 2024]. Reference Source
    1. Alanya-Beltran J: Personalized Learning Recommendation System in E-learning Platforms Using Collaborative Filtering and Machine Learning. s.l., Institute of Electrical and Electronics Engineers Inc.;2024.
    1. Amin S, et al. : An adaptable and personalized framework for top-N course recommendations in online learning. Sci. Rep. 2024;14(1):10382. 10.1038/s41598-024-56497-1 - DOI - PMC - PubMed
    1. Ayouni S, Hajjej F, Maddeh M, et al. : A new ML-based approach to enhance student engagement in online environment. PLoS One. 2021;16:e0258788. 10.1371/journal.pone.0258788 - DOI - PMC - PubMed
    1. Childs-Kean L, Edwards M, Smith MD: Use of Learning Style Frameworks in Health Science Education. Am. J. Pharm. Educ. 2020;84(7):ajpe7885. 10.5688/ajpe7885 - DOI - PMC - PubMed

LinkOut - more resources