Modelling an Adaptive e-Learning System Using LSTM and Random Forest Classification

• Published in: 2018 IEEE Conference on e-Learning, e-Management and e-Services (IC3e) pp. 29 - 34
• Main Authors: Anantharaman, Harish, Mubarak, Abdullah, Shobana, B.T
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2018
• Subjects: Adaptation models; Adaptive systems; Bayes methods; Data structures; Dynamism in e-Learning; Electronic learning; Graph Theory; Learning systems; Long Short Term Memory(LSTM); Predictive models; Random Forest classification
• DOI: 10.1109/IC3e.2018.8632646

E-learning is the most sought after mode of self-learning amongst students and learners today. There is an undisputed requirement for dynamism and customized learning materials that augment a student's understanding of the subject. This paper proposes a comprehensive model based on deep learning by enabling the use of Long Short-Term Memory(LSTM) and Random Forest classification, thereby providing a highly personalized e-Learning system. The adaptive learning system would make use of a bidirected graph data structure that would comprise of various levels and phases. The structure and traversal of the graph are determined by the appropriate machine learning and deep learning algorithms. Convolutional Neural Networks(CNN) and Deep learning algorithms such as Long Short-Term Memory(LSTM) help in the identification of the learner style in accordance to the Felder and Silverman learning style modelling(FSLSM).Once the learner style has been determined, the model uses a Random Forest classifier that takes in a number of parameters including the assessment details of the students to return a prediction of the learner level which pertains to the difficulty of the course. The learner style and level, together would determine the adaptivity of the course. In such a mechanism, the e-Learning course would be adapted with respect to the abilities of the learner, thus allowing a superlative understanding of the course.