Clustering Artikel pada Portal Berita Online Menggunakan Metode K-Means

Abstract

The news categories on news portals are so diverse that the performance of the editors is increasing. The number of news articles each month, adds to the editor's task to manually categorize articles into predetermined categories. Clustering can be used to group data so that later it can group data in the same category with similar data. K-Means is a method that can be used to perform clustering. K-Means is a distance-based clustering technique that is divided into a series of clusters and only works for numeric attributes. The K-Means test conducted in this study is intended to compare cluster values. The K-Means made in this study apply TF-IDF, feature selection, and PCA. The cluster value assessment process uses visualization in the form of a bar plot of each metric value that is considered, namely the mean silhouette, accuracy, precision, recall, F1-score, and silhouette score. The results of the research that has been carried out by the K-Means method can achieve 94.93% accuracy and recall, 95.07% precision, and 94.94% F1-score.