Implementasi K-Means Berbasis Graph Untuk Pengelompokan Peneliti Pada Sistem Repositori Riset ITS

Bilfash, Hifnie (2016) Implementasi K-Means Berbasis Graph Untuk Pengelompokan Peneliti Pada Sistem Repositori Riset ITS. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img]
Preview
Text
5113100091-Undergraduate-Theses.pdf

Download (6MB) | Preview

Abstract

Saat ini ITS sudah memiliki Sistem Repositori Peneliti (resits.its.ac.id) selanjutnya disebut resits. Resits merupakan sistem informasi yang secara khusus menyediakan informasi kepada masyarakat seputar dunia penelitian yang ada di ITS. Pada resits terdapat graf kerjasama peneliti ITS yang hubungan antar graf nya merupakan pengolahan kemiripan topik antar peneliti. Graf kerjasama peneliti yang dimiliki resits dapat di terapkan metode clustering, untuk mendapatkan fokus topik penelitian per cluster. Metode clustering yang digunakan adalah K-Means berbasis graf. Langkah pertama dari K-Means berbasis graf adalah membentuk minimum spanning tree dari graf kerjasama peneliti ITS, setelah itu tree yang terbentuk diterapkan threshold untuk mendapatkan initial centroid dari cluster. Selanjutnya initial centroid yang terbentuk digunakan untuk mengelompokkan peneliti yang belum memiliki cluster dengan melakukan traverse tree ke initial centroid terdekat. Disamping itu, sebagai pembanding metode cluster kerjsama dengan K-Means berbasis graf. Pada tugas akhir ini juga dikembangkan pembentukan graph kerja sama dengan ekspansi sinonim kata. Sinonim tersebut nantinya akan digunakan dalam membentuk graph kerja sama peneliti ITS. Metode K-Means berbasis graf memiliki nilai Dunn Index yang lebih baik dibandingkan K-Means berbasis graf ekspansi sinonim kata. Hal ini disebabkan ekspansi sinonim kata pada judul penelitian menyebabkan keunikan pada judul penelitian semakin berkurang. Sehingga pembentukan graf kerjasama ekspansi sinonim kata tidak menghasilkan cluster graf yang lebih baik daripada graf cluster kerjasama peneliti tanpa ekspansi sinonim kata. ====================================================================================== Currently ITS already have Research Repository System (resits.its.ac.id) and referred as resits. Resits is an information system that specifically provide information to the public about research in ITS. In resits there is researchers ITS cooperation graph, the relations between vertex(researchers) is topic similarity between researchers. Researcher cooperation graf can be applied to clustering methods, the purpose is to get the focus topics per cluster. Clustering method used in this final project is graph based KMeans. The first step of the graph based K-Means is forming a minimum spanning tree from ITS research cooperation graph, then the tree that formed applied threshold value to obtain initial centroid of the cluster. Initial centroid subsequently formed is used to classify researchers who do not have cluster by traversing the tree to the nearest centroid. In addition, as a comparison method of forming the cooperation of researchers graph described above. In this final project also developed the establishment graph of cooperation with expand synonym of a word. The synonyms will be used in the form of ITS research cooperation graph. Graph based K-Means method have better Dunn Index values than graph based K-Means expand synonim. this is due to the expansion synonym for the title of research led to reduces the x uniqueness of the research title. So that graph based K-Means expand synonim method not produce better graph cluster than graph based K-Means without expand synonim.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Sistem Repositori Peneliti; Sinonim; Graph; Kerja sama; Teks Processing; Research Repository System; Synonym; Colaboration; Text Processing
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science. EDP
Q Science > QA Mathematics > QA278 Cluster Analysis. Multivariate analysis
Divisions: Faculty of Information Technology > Informatics Engineering > (S1) Undergraduate Theses
Depositing User: - HIFNIE BILFASH
Date Deposited: 07 Apr 2017 06:30
Last Modified: 26 Dec 2018 06:56
URI: http://repository.its.ac.id/id/eprint/2987

Actions (login required)

View Item View Item