Margaret Alexis L. Oquias, Angie M. Ceniza-Canillo, Riane Nichelle Daeve I. Acharon
Fake news is proven to be rampant in all areas, whether it is in a non-digital outlet, like newspapers, magazines, or gossip from one person to another. This is not to say the digital spread of news is accurate. In fact, it is far from it; fake news in digital media is much more common. It is especially evident in the new pandemic, at a time when people are in need of accurate information about the effects, tracing, and, most importantly, the progress of the vaccine for COVID-19. The purpose of the study was to create a model that would help internet users to separate fake news from real through its identification in several news articles focusing on topics relating to the COVID-19 vaccine. In order to aid in this issue, the researchers intended to use machine learning to create a classification model to identify whether articles in particular news sites, both reliable and unreliable, were considered to be fake news or not. This was done through the analysis of a given dataset and comparing these datasets with another similar dataset created by the researchers and using existing classification algorithms in order to identify which of them would serve the best results. After which, certain metrics like accuracy, precision, recall and the like were employed to evaluate the created model. The results at the end of the study revealed that news articles on both the gathered reliable and unreliable news sites were mostly positive, and that the best classification algorithm to use for the created model is logistic regression with an accuracy of 83.50%, a precision of 82.37%, a recall of S3.50%, a ROC of 95.99%, and an F-Measure of 82.54%. With these results, the researchers hope to highly contribute to the reduction of fake news on the internet for the benefit of the common good. © 2022 IEEE.
Sciences and Mathematics University of San Carlos, Department of Computer Information, Cebu City, Philippines
