Developing a Personalized E-Learning and MOOC Recommender System in IoT-Enabled Smart Education

Abstract

Smart strategies and intelligent technologies are enabling the designing of a smart learning environment that successfully supports the development of personalized learning and adaptive learning. This trend towards integration is in line with the growing prevalence of Internet of Things (IoT)-enabled smart education systems, which can leverage Machine Learning (ML) techniques to provide Personalized Course Recommendations (PCR) to students. Current recommendation techniques rely on either explicit or implicit feedback, often failing to capture changes in learners' preferences effectively as they integrate both types of feedback. This paper proposes a new model for personalized learning and PCR that is enabled by a smart E-Learning (EL) platform. The model aims to gather data on students' academic performance, interests, and learning preferences, using this information to recommend the most beneficial courses for each student. Our approach suggests courses based on the learner's interactions with the system and the cosine similarity of related content, combining explicit (user ratings) and implicit (views and behavior) methodologies. The method employs various ML algorithms and an EL Recommender System (RecSys) based on Collaborative Filtering (CF), including Random Forest Regressor (RFR), Decision Tree Regressor (DTR), K-Nearest Neighbors (KNN), Singular Value Decomposition (SVD), eXtreme Gradient Boosting Regressor (XGBR), and Linear Regression (LR). To evaluate our proposed solution, we benchmark it against existing approaches in terms of predictive accuracy and running time. Experiments are conducted using two benchmark datasets from Coursera and Udemy. The proposed model outperforms existing top-K recommendation techniques in terms of accuracy metrics such as precision@k, Mean Average Precision (MAP)@k, recall@k, Normalized Discounted Cumulative Gain (NDCG)@k, Mean Squared Error (MSE)@k, Root Mean Squared Error (RMSE)@k, and Mean Absolute Error (MAE)@k for PCR. The results show that SVD performs particularly well, demonstrating higher precision, recall, MAP, and NDCG along with lower MAE, RMSE, and MSE values compared to other proposed algorithms. This success can be attributed to SVD's ability to capture complex interactions between students and courses. Our proposed solutions exhibit promise across two datasets and can be applied to various RecSys domains.