Klasifikasi Kategori Dan Pelabelan Berita Bahasa Indonesia Menggunakan Mutual Information Dan K-Nearest Neighbors

Abstract

Along with the increasing development of the internet, the growth of textual information on the internet continues to increase. Increased dissemination of information causes the news released every day to increase. The large number of news, especially those in online media, raises problems in categorizing existing news topics. So we need a system that can categorize every news topic that is in online media and also labeling sentiment on a news. This aims to be able to monitor a situation such as political, social or economic from online news by utilizing classifications and also sentiment labeling so that it can predict steps that can be taken in the future. The method used in this research is feature selection and the KNN classification algorithm with Manhattan distance. In this study, there are two main functions of the system, namely news categorization and also news sentiment labeling. The research stages began with data collection, preprocessing, mutual information feature selection and classification. The results of this study show positive results because it can recognize news categorization and news sentiment labeling.