ZULKARNAEN, FAHRI IZZUDDIN (2023) Analisis Sentimen Tokoh Politik Sebagai Kajian Kandidat Capres Dan Cawapres Menggunakan Fasttext Embeddings Dan Convolutional Neural Network Pada 10 Wilayah Di Indonesia. Undergraduate thesis, UNIVERSITAS PEMBANGUNAN NASIONAL "VETERAN" JAWA TIMUR.
|
Text (Cover)
19081010046-cover.pdf Download (657kB) | Preview |
|
|
Text (Bab 1)
19081010046-bab1.pdf Download (119kB) | Preview |
|
![[img]](http://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Bab 2)
19081010046-bab2.pdf Restricted to Registered users only until 1 June 2025. Download (957kB) |
|
|
Text (Bab 3)
19081010046-bab3.pdf Restricted to Registered users only until 1 June 2025. Download (650kB) |
|
|
Text (Bab 4)
19081010046-bab4.pdf Restricted to Registered users only until 1 June 2025. Download (1MB) |
|
|
Text (Bab 5)
19081010046-bab5.pdf Download (107kB) | Preview |
|
|
Text (Daftar pustaka)
19081010046-daftarpustaka.pdf Download (191kB) | Preview |
Abstract
The internet has become a means for storing and accessing various information publicly. The community itself tends to use social media to express their opinion on a product, political policy, and politicians both party and personal. Sentiment analysis for presidential and cawapres candidates was carried out with the aim of gaining an understanding of the public's view of each candidate. Based on a qualitative analysis of the popularity and sentiments of the Indonesian people on social media, we can identify the names of candidates who have the potential to excel as candidates for president or vice president in the presidential election, which is still more than a year away. Sentiment analysis is done by retrieving data from social media Twitter. This study divides sentiment into 4 classes, namely happy, sad, love, angry. Labeling was done using a pre-trained Roberta model. Modeling is done by comparing the performance of FastText embeddings, Keras embeddings and FastText average word vectors using a convolutional neural network algorithm. Comparisons are made by looking at the evaluation matrix of the best value generated by each model. As a result, FastText embeddings excel with an f1-score value of 0.9510, Keras embeddings has the best f1-score value of 0.9443 and FastText average word vector with an f1-score value of 0.7861. Sentiment calculation is carried out by calculating based on points. The calculation of these points includes happy 2 points, love 1 point, sad -1 point, angry -2 points. The results obtained by Ganjar Pranowo excelled in 9 divisions of the region with a total of 42,216 points, making Ganjar Pranowo a strong candidate to become a presidential candidate. In the vice presidential candidate himself, Erick Thohir excelled in becoming a strong candidate as cawapres by excelling in 5 regional divisions with a total of 55,464 points obtained.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | T Technology > T Technology (General) | ||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
| Depositing User: | Fahri Izzuddin Zulkarnaen | ||||||||||||
| Date Deposited: | 06 Jun 2023 05:17 | ||||||||||||
| Last Modified: | 06 Jun 2023 05:17 | ||||||||||||
| URI: | http://repository.upnjatim.ac.id/id/eprint/14355 |
Actions (login required)
 |
View Item |