Analisis Sentimen terhadap Rencana Pembangunan Ibu Kota Nusantara (IKN) Berdasarkan Twitter (X) Menggunakan Metode Hybrid RoBERTa-GRU

Deagusti, Chika Ananda (2024) Analisis Sentimen terhadap Rencana Pembangunan Ibu Kota Nusantara (IKN) Berdasarkan Twitter (X) Menggunakan Metode Hybrid RoBERTa-GRU. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002201107-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (15MB)

Abstract

Pemindahan ibu kota negara dari DKI Jakarta ke Ibu Kota Nusantara (IKN) di Pulau Kalimantan telah menjadi topik perdebatan yang ramai di masyarakat. Twitter, atau X, telah menjadi platform populer untuk menyampaikan pendapat dan aspirasi masyarakat terkait isu-isu terkini. Penelitian ini bertujuan untuk melakukan analisis sentimen terhadap opini masyarakat di Twitter terkait rencana pembangunan IKN. Dengan analisis sentimen, dapat diidentifikasi dinamika perasaan masyarakat terhadap rencana tersebut. Penelitian ini menggunakan metode hybrid pendekatan Transformer, khususnya Robustly optimized BERT pretraining approach (RoBERTa), dengan metode Recurrent Neural Network (RNN) berupa Gated Recurrent Unit (GRU). Data yang digunakan merupakan data dari Twitter (X) yang berisi pendapat dan sentimen masyarakat terkait rencana pembangunan IKN. Proses analisis dilakukan melalui tahap pengumpulan data, pre- processing data yang mencakup analisis tanda baca, pembersihan data, casefolding, tokenizing, normalisasi, dan pelabelan data menggunakan InSet Lexicon. Data kemudian dibagi menjadi data train, data validation, dan data test sebelum dilanjutkan ke tahap penerapan metode klasifikasi dengan metode hybrid RoBERTa-GRU, serta visualisasi dengan word cloud. Dengan menerapkan metode hybrid RoBERTa-GRU, penelitian ini mendapatkan hasil analisis sentimen Twitter dari Juni 2023 hingga Januari 2024 didominasi dengan sentimen negatif sebanyak 7.907 tweet, dibandingkan dengan 7.126 tweet yang bersentimen positif. Kinerja model hybrid RoBERTa-GRU terbaik didapatkan menggunakan skenario set C dengan parameter optimizer AdamW, batch size 32, GRU hidden size 64, learning rate 1e-7, weight decay 0.08 dan menjalankan 23 epoch mendapatkan loss pengujian sebesar 0.6002 dan akurasi pengujian sebesar 72.78%. Metrik evaluasi lainnya seperti presisi, recall, dan f1-score masing-masing yaitu 72.65%, 69.64%, dan 71.06%
========================================================================================================================================
The relocation of the nation's capital from DKI Jakarta to the Indonesian Capital City (IKN) on the island of Borneo has become a hot topic of debate in society. Twitter, or X, has become a popular platform for conveying public opinions and aspirations regarding current issues. This study aims to conduct a sentiment analysis of public opinion on Twitter regarding the IKN development plan. With sentiment analysis, the dynamics of public feelings towards the plan can be identified. This study uses a hybrid Transformer approach method, specifically the Robustly optimized BERT pretraining approach (RoBERTa), with the Recurrent Neural Network (RNN) method in the form of a Gated Recurrent Unit (GRU). The data used is data from Twitter (X) which contains public opinions and sentiments regarding the IKN development plan. The analysis process is carried out through the stages of data collection, data pre-processing which includes punctuation analysis, data cleaning, case-folding, tokenizing, normalization, and data labeling using InSet Lexicon. The data is then divided into train data, validation data, and test data before continuing to the stage of applying the classification method with the hybrid RoBERTa-GRU method, as well as visualization with word clouds. By applying the hybrid RoBERTa-GRU method, this study obtained the results of Twitter sentiment analysis from June 2023 to January 2024 dominated by negative sentiment of 7,907 tweets, compared to 7,126 tweets with positive sentiment. The best performance of the hybrid RoBERTa-GRU model was obtained using the C set scenario with AdamW optimizer parameters, batch size 32, GRU hidden size 64, learning rate 1e-7, weight decay 0.08 and running 23 epochs obtained a test loss of 0.6002 and a test accuracy of 72.78%. Other evaluation metrics such as precision, recall, and f1-score were 72.65%, 69.64%, and 71.06%, respectively

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Analisis Sentimen, GRU, Ibu Kota Nusantara (IKN), RoBERTa, Twitter;
Subjects:	Q Science > QA Mathematics Q Science > QA Mathematics > QA76.6 Computer programming. Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Chika Ananda Deagusti
Date Deposited:	12 Aug 2024 06:34
Last Modified:	27 Aug 2024 06:47
URI:	http://repository.its.ac.id/id/eprint/115153

Actions (login required)

View Item