Kurniawan, Taufik (2017) Implementasi Text Mining pada Analisis Sentimen Pengguna Twitter Terhadap Media Mainstream Menggunakan Naive Bayes Classifier dan Support Vector Machine. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
1313100075-Undergraduate_Theses.pdf - Accepted Version Download (1MB) | Preview |
Abstract
Keresahan masyarakat terhadap pemberitaan media mainstream sempat menjadi trending topic di media sosial Twitter akibat ketidakpuasan terhadap media yang dinilai tidak representatif dan independen dalam memuat berita. Bahkan pemerintah membahas secara khusus terkait bahaya berita palsu yang sering beredar. Beberapa media mainstream yang fokus sebagai media berita dan banyak mendapat tanggapan masyarakat di media sosial adalah TV One, Metro TV, dan Kompas TV. Sehingga perlu dilakukan penelitian guna mengetahui bagaimana sentimen publik terhadap ketiga media tersebut, apakah mayoritas publik menilai positif atau negatif. Tanggapan publik mengenai media mainstream didapat dari Application Programming Interface (API) pada Twitter karena media sosial tersebut memiliki pengguna yang sangat banyak di Indonesia bahkan hingga mencapai 19,5 juta pengguna dari total 300 juta pengguna global. Pada penelitian ini, praproses teks yang digunakan adalah case folding, tokenizing, stopwords, dan stemming. Untuk praproses stemming digunakan algoritma confix-stripping stemmer Sedangkan pada analisis klasifikasi data teks tersebut digunakan metode Naïve Bayes Classifier dan Support Vector Machine. Klasifikasi menggunakan NBC pada data media TV One dan Kompas TV menghasilkan akurasi sebesar 95,6% dan 97,8%, sedangkan pada media Metro TV menghasilkan nilai G-mean dan AUC berturut-turut sebesar 81,3% and 82,36%. Klasifikasi menggunakan SVM pada data media TV One dan Kompas TV menghasilkan akurasi sebesar 97,9% dan 99,3%,, sedangkan pada media Metro TV menghasilkan nilai G-mean dan AUC berturut-turut sebesar 97,35% and 97,38%.
==================================================================
Public unrest on mainstream media had become a trending topic on Twitter. This situation happened due to the public dissatisfaction on the media that is not representative and independent in presenting news. Even the government discussed specifically related to the danger of false news that is often circulated. Some mainstream media that focus as news media and get a lot of public response in social media is TV One, Metro TV, and Kompas TV. So it is necessary to do research to find out how public sentiments to these media, whether the majority of public rate positive or negative. The public response to mainstream media is derived from the Application Programming Interface (API) on Twitter because the social media has a very large number of users in Indonesia even up to 19.5 million users out of a total of 300 million global users. In this research, the text preprocess used is case folding, tokenizing, stopwords, and stemming. For stemming process used confix-stripping stemmer algorithm, while in the text data classification analysis used Naïve Bayes Classifier and Support Vector Machine method. Classification using NBC on TV One and Kompas TV data resulted accuracy about 95,6% and 97,8%, whereas on Metro TV data yielded G-mean and AUC respectively about 81,3% and 82,36%. Classification using SVM on TV One and Kompas TV data resulted in accuracy about 97,9% and 99,3%, whereas in Metro TV data yielded G-mean and AUC value respectively about 97,35% and 97,38%.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Uncontrolled Keywords: | Media mainstream, Naïve Bayes Classifier, Support Vector Machine, Text mining, Twitter |
| Subjects: | H Social Sciences > HA Statistics Q Science > Q Science (General) |
| Divisions: | Faculty of Mathematics and Science > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | Taufik Kurniawan |
| Date Deposited: | 15 Aug 2017 03:33 |
| Last Modified: | 05 Mar 2019 03:42 |
| URI: | http://repository.its.ac.id/id/eprint/48557 |
Actions (login required)
 |
View Item |