Hoax News Detection in Indonesian Political Headlines Using Multinomial Naive Bayes

Social media is a means of online social interaction on the Internet, where users can freely share information. Because of the freedom, it cannot be denied that some people will misuse social responsible for misusing social media as a place to spread false news. Based on a survey of 2,032 respondents conducted by DailySocial.id in 2018, it was concluded that the majority of Indonesians do not have the ability to detect hoax news. Therefore, the research aims to design and build a hoax news detection application using the Android-based Multinomial Naive Bayes algorithm. At the design stage, the application is designed to receive input in the form of textual political news headlines. It then uses the Multinomial Naive Bayes algorithm to detect hoaxes by comparing the resulting text with data sets. In the testing phase, the algorithm is tested on a confusion matrix and shows the degree of hoax detection. The accuracy of the hoax detection is 88.9%, the precision is 93.33%, the recall is 84%, the recall is 84%, and the F1 score is 88.4%. With a detection application, it is hoped that this hoax news will be able to contribute to the online environment of the Indonesian people by verifying the information before sharing it on social media.