Enhancing Personalized Learning of students through Deep Learning in an Adaptive Learning Environment A dissertation submitted for the Degree of Master of Business Analytics

Abstract

ABSTRACT Adaptive learning and online education have gained significant importance in recent years. In this study, we apply Graph Neural Network (GNN) and Recurrent Neural Network (RNN) techniques to the task of Knowledge Tracing. Unlike previous literature that necessitates searching for the most relevant questions, our methodology focuses on the utilization of the most recent questions from the exercise history. Additionally, while prior studies have employed bidirectional graphs to incorporate question information and learning objectives, our model constructs directional graphs that consider the hierarchy of learning objectives. This hierarchical structure guides the propagation of question and learning objective embeddings, enabling a more contextually informed representation of student knowledge. We compare the performance of our model with question embeddings to a model without, revealing that the incorporation of question embeddings significantly enhances predictive accuracy. Our findings underscore the importance of adaptive learning methodologies in online education, offering insights into more effective knowledge assessment and personalized learning experiences