Hoax Detection on Indonesian Tweets using Naïve Bayes Classifier with TF-IDF

Twitter is one of the most popular social media platforms in world nowadays. users Indonesia are fifth largest and always active expressing themselves getting information through tweets. A hoax a lie created as if it were true. Hoaxes also often spread via The hoaxes extremely dangerous because can cause discord even misunderstanding. Therefore, must be resisted. This study aims to build system detect on Indonesian objective this research identify tweets by using Naïve Bayes classifier with Term Frequency Inverse Document (TF-IDF). collects annotates from post which sent user account. applied several text preprocessing techniques provide datasets. To best prediction model, work splits datasets into training testing There four experimental scenarios that refer splitting dataset. results showed model TF-IDF had 64% accuracy recall, 69% 67% precision, F1-score respectively. result superior when without TF-IDF. It means has made positive contribution improving performance. Finally, contributes detecting news proclivity for filtering what classified or not.