Improving The Accuracy of Naïve Bayes Algorithm for Hoax Classification Using Particle Swarm Optimization

摘要

Hoax news circulation is very widespread which occurs in the media information, both print and online media. For some people hoax news can only appear on the online media. But printed medias also often include hoax news in their published news. In the present era, it is very important providing information with relevant and additional facts otherwise it is categorized as hoax. Therefore, hoax classification approach is needed. This paper focuses on improving the accuracy of hoax classification in textual documents contents. Naive Bayes algorithm is used to train dataset with the use of PSO in the algorithm. Experiment is conducted with the trained model over 600 documents. It shows that feature selection with PSO affects the classification results performed using Naïve Bayes. Accuracy increased from 91.17% without using feature selection, to 92.33% when feature selection is carried out using PSO.