Hoax Identification of Indonesian Tweeters using Ensemble Classifier

Fake information, better known as hoaxes, is often found on social media. Currently, media not only used to make friends or socialize with online, but some use it spread hate speech and false information. Hoaxes are very dangerous in life, especially countries large populations ethnically diverse cultures, such Indonesia. Although there have been many studies detecting the accuracy efficiency still need be improved. To help prevent of these we built a model identify information Indonesian using an ensemble classifier that combines n-gram method, term frequency-inverse document frequency, passive-aggressive method. The evaluation process was carried out 5000 samples from Twitter accounts this study. testing four schemes by dividing dataset into training test data based ratios 90:10, 80:20, 70:30, 60:40. inspection results show our software can accurately detect hoaxes at 91.8%. We also increase precision hoax detection proposed method compared several previous studies. developed various platforms.