Clustering Dokumen Skripsi Berdasarkan Abstrak dengan Menggunakan Bisecting K-Means

Abstract

The need of thesis data searching increases every year along with the increase in the number of students. Search of reference by tracing documents one by one takes a lot of time. Therefore, a system that is capable of clustering documents automatically is necessary. This study developed a system to perform clustering of theses automatically based on their abstracts. It used bisecting Kmeans method to cluster the data. The data in this research were from IPB’s Computer Science bachelor theses, comprising 78 abstracts in Indonesian and 113 abstracts in English. The result showed that clustering the documents using bisecting K-means could be done with the best value of i threshold (internal cluster distance) of 0.67 for the Indonesian abstracts resulting in a rand index of 0.867, while the best i threshold value for the English abstracts was 0.55 resulting in a rand index of 0.862.