Support Vector Machine For Hoax Detection

Abstract

Along with the development of information technology, news media has also developed by presenting information online Along with the rapid development of online news, the spread of fake news information (hoaxes) is also increasing rapidly and widely. Hoax news is often spread intentionally for various purposes. Generally, hoax news aims to direct the reader's perception to believe in a bad perception of an event, character or even a company. The motivation is to invite readers to believe something that is not true with the aim of benefiting the news disseminator is something dangerous. This research aims to detect English-language hoaxes by applying the Support vector machine (SVM) algorithm. In this study, the data used are two data sources, namely English news datasets from Kaggle and English news taken from BBC. The results of this study show that the application of the SVM algorithm turns out to get good performance because the model is able to classify hoax news with an accuracy of 99.4% on Kaggle data while on the BBC news dataset the model gets an accuracy of 98.9%. This research also shows that the SVM method is proven to have good generalization properties. Where it is able to identify test data that is completely different from the training data.