Swanjaya, Daniel (2015) Hypergraph-Partitioning Pada Co-Authorship Graph Untuk Pengelompokan Penulis Berdasarkan Topik Penelitian. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
5112201022-Master Thesis.pdf - Published Version Download (1MB) | Preview |
Abstract
Topik Penelitian dapat diketahui dari Abstraksi dokumen penelitian,
misalnya laporan Karya Tulis Ilmiah (KTI) berupa Tugas Akhir, Tesis dan
Disertasi. Topik Penelitian dari KTI merupakan kumpulan kata-kata penting yang
menunjukkan area/bidang penelitian dari KTI tersebut. Sebuah KTI dibimbing
beberapa Dosen pembimbing, dan seorang Dosen biasanya akan membimbing
beberapa topik tertentu. Beberapa Dosen yang memiliki bidang penelitian yang
sama membentuk grup riset dalam lingkup Jurusan, tetapi beberapa Jurusan
terdapat Dosen yang memiliki kesamaan atau kemiripan bidang penelitian. Pada
Tesis ini diusulkan metode untuk mengelompokkan Penulis (Dosen) berdasarkan
kesamaan topik penelitian pada Co-Authorship Graph menggunakan Hypergraph
Partitioning, sehingga memungkinkan untuk membuat grup riset dalam lingkup
antar Jurusan atau tingkat perguruan tinggi. Metode dibagi menjadi empat tahap
yaitu praproses, ekstraksi topik penelitian, pembentukan Co-Authorship Graph, dan
pengelompokan Penulis. Ekstraksi topik penelitian, mendapatkan topik dari KTI
berdasarkan Judul dan Abstraksi menggunakan Latent Dirichlet Allocation (LDA).
Pembentukan Co-Authorship Graph, dimana node adalah Penulis, edge adalah
hubungan kolaborasi dan kesamaan/kemiripan topik penelitian, dan bobot edge
adalah nilai jaccard dan cosine similary topik penelitian antar Penulis.
Pengelompokkan Penulis pada Co-Authorship Graph menggunakan Hypergraph
Partitioning. Uji coba metode menggunakan data Penelitian dari Lembaga
Penelitian dan Pengabdian Masyarakat (LPPM) ITS Surabaya. Hasil
pengelompokkan divalidasi menggunakan Silhouette dan Entropy. Hasil akhir
pengelompokkan menunjukkan bahwa telah terbentuk kelompok Penulis yang
anggotanya berasal dari Jurusan atau bidang yang berbeda, dengan kesamaan topik
yang tinggi.
===============================================================================================
Research topics are able to be known from the abstract section of research
document, for example the document of literature, final project, thesis and
dissertation. Research topics from literature are group of important word which are
referred to the field of its literature. A literature is guided by some advisors and an
advisor will handle some research topics. Advisors who have same research topics
will form a research group in the faculty. In this thesis, a method is proposed to
form group of researchers based on their topics with Co-Authorship Graph, it uses
Hypergraph-Partitioning so it will possible to form research group in faculty or
university. The method is divided to four phases, pre-processing, extracting
research topic, forming Co-Authorship Graph and dividing authors. Research topic
extraction, the topic is obtained from the title and the abstract using Latent Dirichlet
Allocation (LDA). Co-Authorship Graph is formed, the node is author, the edge is
collaboration and similarity research topic, and node’s weight is jaccard value and
cosine similarity of researcher topic. The authors agglomeration in Co-Authorship
Graph use Hypergraph Partitioning. The testing method use the data from Lembaga
Penelitian dan Pengabdian Masyarakat (LPPM) ITS Surabaya. The result is
validated by silhouette and entropy and it shows the agglomeration of authors. The
member is obtained from the faculty or different field with high similarity topics.
| Item Type: | Thesis (Masters) |
|---|---|
| Additional Information: | RTIf 004.22 Swa h |
| Uncontrolled Keywords: | Graph Clustering, Latent Dirichlet Allocation, Co-Authorship Graph, Hypergraph Partitioning. |
| Subjects: | T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques |
| Divisions: | Faculty of Information Technology > Informatics Engineering > 55101-(S2) Master Thesis |
| Depositing User: | Mr. Tondo Indra Nyata |
| Date Deposited: | 19 Jun 2017 06:59 |
| Last Modified: | 24 Aug 2018 07:05 |
| URI: | http://repository.its.ac.id/id/eprint/41741 |
Actions (login required)
 |
View Item |