Clustering Tagg Status Facebook Dengan Menggunakan Algoritma K-MEDOIDS

Abstract

This research is implementing K-Medoids algorithm to discover clusters on a friend list of a Facebook user. To find those clusters, the system uses the strongest path which is based on the tag frequency of status update of the facebook user to measure the tie strength from a friend to other friends. The experiments of using 3 clusters, 5 clusters, and 7 clusters, which resulted in average purity score 0.7430. The experiment resulted in rank of highest average purity score, at the first rank is experiment which used 3 clusters with the average score 0.8806, at the second rank is experiment which used 7 clusters with the average score 0.7114, and the third rank is experiment which used 5 clusters with the average score 0.6368.

Keywords: cluster, Dijkstra, Facebook, strongest path, K-Medoids, purity, status update, tag