Data Augmentation for Hoax Detection through the Method of Convolutional Neural Network in Indonesian News

Abstract

The concept of hoax or fake news refers to the intentional spread of false information on social media that aims to confuse and mislead readers to achieve an economic or political agenda. In addition, the increasingly diverse and numerous actors in the field of news writing and dissemination have led to the creation of news articles that need to be recognized whether they are credible or not. Furthermore, hoax can harm the social and political aspects of Indonesian society. Central Connecticut University released a study entitled The World's Most Literate Nations in 2016, where Indonesia ranked 60th out of 61 countries, indicating that Indonesian media literacy still needs to improve in critically evaluating information and distinguishing between fake news and valid news. Based on this description, the research will create the Synonym-Based Data Augmentation for Hoax Detection using the Convolutional Neural Network (CNN ) method and Easy Data Augmentation (EDA). This research resulted in an accuracy of 8,.81, indicating that it can be stated to be accurate in detecting hoax news