Improving the educational experience on Youtube: a machine learning approach to classifying and recommending educational videos

Abstract

The fast development of technology has revolutionized social interaction and enabled easy access to a vast amount of information. However, it is increasingly challenging to find relevant educational materials within the large volume of available data. This challenge has led to a significant waste of time for teachers and students in searching for high-quality educational resources. In this sense, the present work focuses on classifying educational videos on YouTube using Machine Learning models. The study extends a previous work that analyzed YouTube videos and proposed a methodology for classifying them using their comments. The current study expands the dataset used in the previous work and employs Machine Learning algorithms such as Random Forest and Neural Networks, along with hyperparameter tuning techniques like Grid Search. Experimental results showed that a Convolutional Neural Network was able to differentiate educational videos from non-educational ones with an accuracy rate of 95,71%. This study highlights the potential of Convolutional Neural Networks in classifying educational content on YouTube, contributing to advances in the field of Machine Learning for educational purposes.