Support Vector Machine-Based Hoax Detection on Indonesian Online News

Abstract

The rapid development of information technology and social media resulted in more information dissemination in the form of digital news and transformed analog media into online media. However, the spread of news in online media is not all true (hoax). The problems faced by internet users of hoax news can be solved through pattern recognition. Research data was conducted of 2,000 Indonesian news, 1,000 for non-hoax news, and 1,000 for hoax news. Data was collected from Indonesian online news portals by crawling text methods. The research model was built using CRISP-DM methodology. The results showed that as high as 96,01% accuracy can be achieved using Support Vector Machine. Based on these results, it can be concluded that this modeling can be used to support the detection of hoax news in Indonesia today. It is hoped that this model can also be applied to help the government filter news that will be distributed to the people of Indonesia.