Ferdiansyah, Irwin Deriyan (2018) Pengelompokan Dokumen Petisi Online Di Situs Change.Org Menggunakan Algoritme Hierarchical Clustering Upgma. Sarjana thesis, Universitas Brawijaya.
Abstract
Change.org merupakan salah satu website yang sering digunakan oleh masyarakat untuk sarana penyampaian petisi dan kampanye sosial secara online. Kampanye lewat media sosial terbukti dapat menghasilkan perubahan. Aliran informasi petisi online yang berupa dokumen diperbarui setiap harinya dalam jumlah yang besar, membuat clustering dokumen menjadi sangat penting. Clustering dokumen adalah proses pengelompokan dokumen yang memiliki kesamaan topik. Tujuannya untuk membagi dokumen berdasarkan kesamaan, sehingga memudahkan dalam proses pencarian. Metode yang digunakan adalah hierarchical clustering UPGMA atau Unweighted Pair-Group Method using Arithmetic averages dengan menambahkan reduksi fitur menggunakan metode Latent Semantic Indexing hasil pemecahan matrik Singular Value Decomposition. Hasil penelitian menyimpulkan bahwa Latent Semantic Indexing dapat mengatasi permasalahan pada data berdimensi tinggi. Data yang digunakan berjumlah 100 petisi. Dari hasil pengujian performansi menggunakan Cophenetic Correlation Coefficient diperoleh nilai cophenetic sebesar 0,75959 pada rank matrik LSI sebanyak 10% dan Silhouette Coefficient sebesar 0,36862 dengan jumlah cluster sebanyak 2 cluster.
English Abstract
Change.org is a website that is often used by people, which means for online delivering petitions and social campaignings. Campaign through social media had been proven that can make a change. The flow information of online petitions documents is updated daily in large numbers. It makes documents clustering being very important. Documents clustering is a process of grouping documents which have same topic. It aims to devide documents by its similarly, so the process of searching will be easier. This study uses hierarchical clustering UPGMA or unweighted pair-group method by arithmetic averages with adding feature reduction using Latent Semantic Indexing method, that is the result of splitting Singular Value Decomposition matrix. The result of this study conclude that Latent Semantic Indexing method can solved the problem in high-dimensional data. The data conducted by 100 petitions. The result of performance testing which used Cophenetic Correlation Coefficient obtained conphenetic value of 0.75959 at LSI matrix rank of 10 % and Silhouette Coefficient of 0.36862 with number of clusters as many as 2 clusters.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | SKR/FTIK/2018/142/051801129 |
| Uncontrolled Keywords: | pengelompokan dokumen, Change.org, UPGMA clustering, Singular Value Decomposition, Latent Semantic Indexing, Silhouette Coefficient, Cophenetic Correlation Coefficient. Document Clustering, Change.org, UPGMA clustering, Singular Value Decomposition, Latent Semantic Indexing, Silhouette Coefficient, Cophenetic Correlation Coefficient. |
| Subjects: | 000 Computer science, information and general works > 020 Library and information sciences |
| Divisions: | Fakultas Ilmu Komputer > Teknik Informatika |
| Depositing User: | Budi Wahyono Wahyono |
| Date Deposited: | 12 Mar 2019 02:22 |
| Last Modified: | 16 Oct 2021 03:48 |
| URI: | http://repository.ub.ac.id/id/eprint/13555 |
Preview |
Text
Irwin Deriyan Ferdiansyah.pdf Download (11MB) | Preview |
Actions (login required)
 |
View Item |