Pengembangan Aplikasi Pendeteksian Hoaks Pada Berita Covid-19 Berbahasa Indonesia Menggunakan Naïve Bayes Bernoulli

Abstract

The main objective of this research is to develop an application that can detect hoax and not hoax at the covid-19 articles with Indonesian language. The dataset was obtained through web scraping method at turnbackhoax.id website on April until November 2020. Then we will perform the labelling of the dataset, which are zero for hoax and one for not hoax. The labelling result of the data obtained is 499 data of COVID-19 articles with Indonesian language. This research is using incremental method as an application development. The incremental method that used is consist into two steps, and those are incremental part one that contains the development of classification model, which are data preprocessing, Bernoulli Naïve Bayes algorithm with TF-IDF as term weighting method, and the testing of classification model with five new testing data. While, incremental part two contains the development of the application user interface. The interface of the application exists two buttons, which are process and exit button. The results of the classification model are 84% of accuracy, 84,15% of precision, 84,0% of recall, and 84,04% of f-measure.