Currently, a lot of news about politics can cause misinterpretation because the news is not necessarily true or a hoax. Hoax is fake news spread to gain trust so that people feel confident the content is true. Another impact of hoaxes can be detrimental to people's emotions and finances. In Indonesia, especially DKI Jakarta, the spread of hoaxes ahead of the general election, between September 2023 and February 2024, is very high. To filter articles on social media, it is necessary to classify political news whether it falls into the hoax or non-hoax category. The hoax classification process uses a preprocessing stage which consists of tokenization and winnowing. Followed by the process of weighting words and Jaccard similarities until the classification process using the N-Gram method can provide accuracy in determining hoax or non-hoax news on social media. The aim of this research is to build a system model that can find political news information. The system will systematically determine news into a predetermined hoax or non-hoax label or category. Using a sample of 200 articles obtained from online social media. The article used is in the form of text without images included. Articles are taken from online social media because a lot of information is spread through social media. The research will be made into two categories or labels, namely hoax and non-hoax. The evaluation results in this research use a confusion matrix, where the precision obtained is 0.90, the recall is 0.78 and the accuracy obtained is 85%. Using the Vector Space Model method and Jaccard stemming, and similarity can make it easier to classify large numbers of articles.
Author
(s) Details
Wiwin Windihastuty
Universitas Budi Luhur, Jl. Ciledug Raya, Pesanggrahan, Jakarta Selatan,
Indonesia.
Goenawan Brotosaputro
Universitas Budi Luhur, Jl. Ciledug Raya, Pesanggrahan, Jakarta Selatan,
Indonesia.
Please see the book here:- https://doi.org/10.9734/bpi/mono/978-93-48859-98-3/CH23
No comments:
Post a Comment