E-Learning Recommendation System Using Enhanced Firefly and Fine-Tuned K-Nearest Neighbor Algorithm

Abstract

Nowadays, the rapid growth of e-learning platforms has led to an overwhelming amount of educational content available online. While this abundance of resources is beneficial, it can also create challenges for learners in identifying the most relevant content tailored to their needs and preferences. To resolve this problem, Recommendation Systems (RS) have become essential tools for e-learning platforms, helping to personalize the learning experience by suggesting courses, articles, and other educational materials that align with users' interests and learning goals. With the application of an improved Firefly Algorithm (FA) and a refined K-Nearest Neighbor (KNN) algorithm, this research aims to create an advanced e-learning Recommendation System (RS). The goal is to use the advantages of both classification and optimization methods to give learners individualized, high-quality recommendations. The methodology begins with pre-processing using K-Means Clustering (KMC) to effectively handle noise and identify dense data regions. Subsequently, an Enhanced Firefly Algorithm optimizes system parameters, achieving optimal fitness values (FV) for recommendation accuracy. The integration of Content-Based and Collaborative Filtering techniques by the RS to extract features and insights from data. Content-Based Filtering (CBF) by Cosine similarity (CS) focuses on item similarities and user preferences derived from item attributes, while Collaborative Filtering by Improved genetic algorithm (IGA) leverages user interactions to predict preferences based on similar users or items. These filtering methods are complemented by a classification approach using a fine-tuned K-Nearest Neighbour (KNN) algorithm adjusted by Artificial Neural Networks (ANN). The prediction accuracy, precision, recall, and reduces mean absolute error (MAE) was enhanced by this hybrid approach.