Machine Learning Framework for Assessing Digital Literacy in Higher Education

Abstract

Digital literacy in higher education is essential for lifelong learning, as technology has swiftly transformed communication, work, and study practices. Schools and colleges are having a hard time keeping up with the rapid growth of employment skills and the need for new ones, especially since that society is going through big changes in technology and social norms. This study compiles educational data from many sources and use data preprocessing to cleanse, arrange, and transform raw data into significant digital literacy indicators for predicting levels of digital literacy in HE. This work uses Latent Dirichlet Allocation (LDA) feature extraction to identify noteworthy patterns and enhance the dataset. Using past training data, a classification method based on Support Vector Machine (SVM) from statistical learning theory predicts digital literacy, which solves difficult prediction problems. The proposed model performs effectively, with the SVM method achieving an optimal accuracy of 95.15%, representing a substantial enhancement above the majority baseline accuracy. These results suggest that LDA and SVM may be used to measure and enhance Digital Literacy in Higher Education, which will help with online learning and learning new skills for life.