Identification and analysis of real and fake news by XGBoost algorithm of machine learning

Abstract

As the internet develops rapidly, fake news has become increasingly easy to propagate. Numerous academics acknowledge the perilous nature of this phenomenon, particularly in the context of the contemporary “post-truth era,” highlighting its substantial risk to the public. Hence, the detection and halting of fake news dissemination are absolutely vital. This study utilizes machine learning’s eXtreme Gradient Boosting (XGBoost) algorithm to construct a model that can differentiate between genuine and fake news. The model is compared with others utilizing different algorithms and is ultimately selected. The study successfully constructs a model with an accuracy rate of approximately 95% in identifying real and fake news. This model provides the public with a convenient way to differentiate between real and fake news and gradually diminishes the threat of fake news. Additionally, this project’s implications extend beyond merely identifying real and fake news. The model can be further developed to detect fake information, providing greater societal benefits.