Klasifikasi Berita Bahasa Indonesia Dengan Menggunakan Metode K-Nearest Neighbor Dan Naive Bayes

Abstract

In the era of rapid development of information technology, the need for a news classification system is crucial to manage the increasing volume of information. This study aims to develop a news classification system in Indonesian into five main categories: Politics, Economy, Health, Security, and Poverty. The methods used include the K-Nearest Neighbor (KNN) algorithm and Naïve Bayes. The dataset consists of 2,000 news items obtained from Kaggle, with preprocessing stages including cleaning, tokenizing, normalization, and TF-IDF weighting. The evaluation was carried out through three data sharing scenarios: 70%-30%, 80%-20%, and 90%-10%. The results showed that the KNN algorithm achieved the highest accuracy of 89% in the 80%-20% and 90%-10% scenarios, while Naïve Bayes produced the best accuracy of 78.66% in the 70%-30% scenario. KNN proved to be more reliable for data with balanced category distribution, while Naïve Bayes required further adjustment, especially for underrepresented data categories. This research provides significant contributions to the development of an automatic news classification system, which can be implemented to improve user experience in accessing information.