Abstract
News is generally published by news providers. News providers in publishing news are sometimes not grouped into several news categories. This allows readers difficulty in finding news because the news is not grouped. This study group Krama Alus Javanese news for readers can search for news according to categories with the Hierarchical and K-Means Algorithms. Research data uses news documents from the JOGJATV online news portal. In this study, the data is processed through text preprocessing. Text preprocessing consists of five processes, one of which is stemming. Stemming is used to adapt the Nazief and Adriani algorithms which are adjusted to the rules in Javanese. The method used to process the results of text preprocessing is hierarchical clustering combined with k-means. Hierarchical clustering is used to determine the number of clusters and centroids of each cluster. The results showed that the stemming process required the addition of basic words and stemming rules. While for the clustering results there are 18 clusters. The evaluation of the overall cluster structure with Average Silhouette Width (ASW) shows a value of 0.9142. This value indicates that a cluster has different characteristics from other clusters so that the news documents are in the right group. The clustering results are also validated by Experts with good results, namely 11 clusters that can be labeled while 7 clusters with labels are not specific.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
