PUTRA, ALIEF RAMADHAN DWI (2025) ANALISIS SENTIMEN PUBLIK TERHADAP PERUBAHAN KEPIMPINAN MENGGUNAKAN ALGORITMA LSTM DAN NAIVE BAYES PADA PLATFORM X. S1 thesis, Universitas Mercu Buana Jakarta.
|
Text (HAL COVER)
01 COVER.pdf Download (559kB) | Preview |
|
![[img]](https://repository.mercubuana.ac.id/style/images/fileicons/text.png) |
Text (BAB I)
02 BAB 1.pdf Restricted to Registered users only Download (104kB) |
|
|
Text (BAB II)
03 BAB 2.pdf Restricted to Registered users only Download (296kB) |
|
|
Text (BAB III)
04 BAB 3.pdf Restricted to Registered users only Download (463kB) |
|
|
Text (BAB IV)
05 BAB 4.pdf Restricted to Registered users only Download (279kB) |
|
|
Text (BAB V)
06 BAB 5.pdf Restricted to Registered users only Download (38kB) |
|
|
Text (DAFTAR PUSTAKA)
07 DFATAR PUSTAKA.pdf Restricted to Registered users only Download (173kB) |
|
|
Text (LAMPIRAN)
08 LAMPIRAN.pdf Restricted to Registered users only Download (613kB) |
Abstract
Political sentiment analysis on social media has become a critical tool for understanding public perception during leadership transitions. This study compares the effectiveness of Long Short-Term Memory (LSTM) and Naïve Bayes algorithms in classifying sentiment related to Indonesia’s 2024 leadership change using data from platform X (formerly Twitter). A total of 5,942 Indonesianlanguage tweets were collected and labeled through both lexicon-based and manual annotation. Manual labeling was crucial in capturing nuanced, contextdependent sentiments often missed by automated methods. LSTM was applied for its strength in modeling sequential patterns in text, while Naïve Bayes served as a lightweight probabilistic baseline. Both models were evaluated using accuracy, precision, recall, and F1-score. Results show that LSTM achieved 72.02% accuracy on lexicon-based data and 78.47% on manually labeled data, while Naïve Bayes reached 61.13% and 78.64%, respectively. LSTM offered better generalization across sentiment classes—especially neutral—whereas Naïve Bayes excelled in detecting clearly polarized sentiment. These findings highlight the importance of model selection based on data characteristics and labeling strategy. The study contributes practical insights for political analysts and institutions aiming to monitor digital public opinion and inform evidence-based policymaking. Kata kunci: Sentiment analysis, Long Short-Term Memory (LSTM), Naive Bayes, Leadership, Social Media. Analisis sentimen politik di media sosial telah menjadi alat penting untuk memahami persepsi publik selama masa transisi kepemimpinan. Penelitian ini membandingkan efektivitas algoritma Long Short-Term Memory (LSTM) dan Naïve Bayes dalam mengklasifikasikan sentimen terkait pergantian kepemimpinan Indonesia pada tahun 2024 menggunakan data dari platform X (sebelumnya Twitter). Sebanyak 5.942 tweet berbahasa Indonesia dikumpulkan dan diberi label menggunakan dua pendekatan, yaitu berbasis leksikon dan pelabelan manual. Pelabelan manual terbukti penting untuk menangkap sentimen yang bersifat kontekstual dan halus, yang sering kali luput oleh metode otomatis. LSTM dipilih karena kemampuannya dalam memodelkan pola berurutan dalam teks, sementara Naïve Bayes digunakan sebagai model probabilistik ringan sebagai pembanding. Kedua model dievaluasi menggunakan metrik akurasi, presisi, recall, dan F1-score. Hasil menunjukkan bahwa LSTM mencapai akurasi sebesar 72,02% pada data leksikon dan 78,47% pada data manual, sementara Naïve Bayes mencapai 61,13% dan 78,64% secara berturut-turut. LSTM menunjukkan generalisasi yang lebih baik pada kelas netral, sedangkan Naïve Bayes unggul dalam mendeteksi sentimen yang lebih terpolarisasi. Temuan ini menegaskan pentingnya pemilihan model berdasarkan karakteristik data dan strategi pelabelan. Studi ini memberikan wawasan praktis bagi analis politik dan institusi dalam memantau opini publik digital serta mendukung pengambilan kebijakan berbasis data. Kata kunci: Analisis sentimen, Long Short-Term Memory (LSTM), Naive Bayes, Kepemimpinan, Media Sosial.
Actions (login required)
 |
View Item |
