Analisis Sentimen Berbasis Aspek dan Penggalian Opini terhadap Karakter Calon Pasangan Presiden pada Komentar YouTube menggunakan Model EMC-GCN

Tsaniyah, Nazriyah Deny (2024) Analisis Sentimen Berbasis Aspek dan Penggalian Opini terhadap Karakter Calon Pasangan Presiden pada Komentar YouTube menggunakan Model EMC-GCN. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5026201151_Undergraduate_Thesis.pdf - Accepted Version Download (6MB)

Abstract

Pada tahun 2024, Indonesia memasuki periode politik krusial dengan pelaksanaan pemilihan umum untuk menentukan presiden dan wakil presiden baru. Evaluasi terhadap karakter calon pemimpin cukup penting dalam mempengaruhi tingkat dukungan publik. Tugas akhir ini menganalisis sentimen publik terhadap calon-calon presiden melalui komentar di video YouTube menggunakan model Enhanced Multi-Channel Graph Convolutional Network (EMC-GCN). Data dikumpulkan dari 27 video unggahan kanal berita resmi seperti KOMPASTV dan CNN Indonesia, menghasilkan file CSV dengan kolom ‘publishedAt’, ‘authorDisplayName’, dan ‘textDisplay’. Model EMC-GCN, dilatih dengan f1-score maksimum 61.52% pada sequence length 120, epoch 50, dan batch size 8, menunjukkan f1-score 56.87% pada dataset test. Meskipun efektif dalam memprediksi sentimen positif, model sering mengalami kesulitan dalam membedakan sentimen netral. Hasil analisis menunjukkan dominasi sentimen positif, terutama pada aspek 'ngomong', 'bicara', dan 'debat'. Namun, terdapat variasi signifikan dalam sentimen, dengan Anies Baswedan mendapat sentimen negatif pada aspek 'kerja', Prabowo Subianto pada aspek 'ngomong' dan 'debat', serta Muhaimin Iskandar pada 'jawaban' dan 'ngomong'. Ganjar Pranowo menunjukkan distribusi seimbang antara sentimen positif dan negatif, sedangkan Gibran Rakabuming dan Mahfud MD menunjukkan pola sentimen yang berbeda, dengan Gibran mendapatkan respon positif dominan dan Mahfud MD mengalami sentimen negatif pada aspek tertentu. Analisis ini menyoroti pentingnya media sosial dalam membentuk persepsi publik terhadap calon pemimpin dan memberikan wawasan mendalam tentang evaluasi karakter calon berdasarkan sentimen.
================================================================
In 2024, Indonesia will enter a crucial political period with general elections to determine the new president and vice president. Evaluation of the character of a potential leader is quite important in influencing the level of public support. This final assignment analyzes public sentiment towards presidential candidates through comments on YouTube videos using the Enhanced Multi-Channel Graph Convolutional Network (EMC-GCN) model. Data was collected from 27 videos uploaded by official news channels such as KOMPASTV and CNN Indonesia, producing a CSV file with the columns 'publishedAt', 'authorDisplayName', and 'textDisplay'. The EMC-GCN model, trained with a maximum f1-score of 61.52% on sequence length 120, epoch 50, and batch size 8, shows an f1-score of 56.87% on the test dataset. While effective in predicting positive sentiment, models often have difficulty distinguishing neutral sentiment. The results of the analysis show the dominance of positive sentiment, especially in the 'talk' and 'debate' aspects. However, there were significant variations in sentiment, with Anies Baswedan receiving negative sentiment on the 'work' aspect, Prabowo Subianto on the 'talking' and 'debate' aspects, and Muhaimin Iskandar on 'answers' and 'talking'. Ganjar Pranowo shows a balanced distribution of positive and negative sentiment, while Gibran Rakabuming and Mahfud MD show different sentiment patterns, with Gibran getting a dominant positive response and Mahfud MD experiencing negative sentiment in certain aspects. This analysis highlights the importance of social media in shaping public perceptions of potential leaders and provides deep insight into candidate character evaluations based on detected sentiments.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Karakter calon presiden, Aspect-based sentiment analysis, YouTube, Ekstraksi opini, EMC-GCN, Politik, Character of presidential candidates, Aspect-based sentiment analysis, YouTube, Opinion extraction, EMC-GCN, Politics
Subjects:	J Political Science > J General legislative and executive papers Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Nazriyah Deny Tsaniyah
Date Deposited:	01 Aug 2024 12:36
Last Modified:	01 Aug 2024 12:36
URI:	http://repository.its.ac.id/id/eprint/110947

Actions (login required)

View Item