Detecting Fake News on COVID-19 Vaccine from YouTube Videos Using Advanced Machine Learning Approaches

Abstract

Fake news is considered a massive threat to many internet users, with the heavy usage of social media networks. Many news agencies develop their platforms to publish and share their news articles. Also, an ordinary user on the social media network has an account where the content can be posted and shared. Some users share fake news or rumors to achieve personal goals and benefits. Fake news is considered to be the most visible challenge on social media networks. It creates a threat to individuals and society while creating a negative impact. Many research works tackle this issue using propagation-based, content-based, and meta-data analysis approaches. This book chapter proposes a model to detect fake news about the COVID-19 vaccine on YouTube videos using a sentiment analysis approach through machine learn and deep learning approaches focusing on the Arabic language of middle-east people. The process started with building a dataset through the collected textual data using the comments that were later annotated into two classes. They are fake and real news. Two experiments have been conducted using machine learning classifiers and deep learning models. Through these experiments, the performance level of the model has reached 94% in terms of accuracy. In the deep learning approach, it has reached 99%.