Metode Hibrid untuk Analisis Sentimen di Twitter: Menggunakan Lexicon, BERT, dan LSTM pada Dataset Kenaikan Harga Bahan Bakar Minyak

Hadiono, Aqil Ramadhan (2025) Metode Hibrid untuk Analisis Sentimen di Twitter: Menggunakan Lexicon, BERT, dan LSTM pada Dataset Kenaikan Harga Bahan Bakar Minyak. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5025201261-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (1MB) | Request a copy

Abstract

Penelitian ini memperkenalkan dan mengimplementasikan metode hibrid untuk analisis sentimen pada data Twitter, dengan mengintegrasikan algoritma Machine Learning, Lexicon, BERT (Bidirectional Encoder Representations from Transformers), dan LSTM (Long Short-Term Memory). Latar belakang penelitian ini adalah pentingnya memahami opini dan reaksi masyarakat terhadap isu-isu yang sensitif dan berdampak luas, seperti kenaikan harga BBM. Media sosial, khususnya Twitter, telah menjadi sumber data yang kaya untuk menganalisis sentimen publik, namun sifat data yang tidak terstruktur, dinamis, dan cenderung informal memerlukan pendekatan analisis yang lebih komprehensif dan adaptif. Penelitian ini menerapkan metode pada konteks hasil kenaikan BBM untuk menggali pandangan dan reaksi masyarakat. Pendekatan hibrid ini bertujuan memberikan solusi analisis sentimen yang komprehensif dan adaptif dalam mengatasi tantangan data Twitter yang dinamis, dengan hasil kenaikan BBM sebagai studi kasus utama. Pada Tugas Akhir ini diimplementasikan beberapa metode untuk membandingkan hasil akurasi, yaitu Lexicon, BERT, LSTM, dan Hibrid. Pada Tugas Akhir ini diimplementasikan beberapa metode untuk membandingkan hasil akurasinya. Yaitu Lexicon, BERT, LSTM dan Hibrid. Tahapan dalam membuat Tugas Akhir ini adalah pengolahan data, praproses, pembuatan model dan pengujian model dilakukan dengan menggunakan confusion matrix, kemudian menghitung akurasi, presisi, recall, dan F1-score. Berdasarkan evaluasi, hasil akurasi terbaik terdapat pada metode hibrid 70% training dan 30% testing mendapatkan hasil 85%.
=================================================================================================================================
This research introduces and implements a hybrid method for sentiment analysis on Twitter data, by integrating Machine Learning (ML), Lexicon, BERT (Bidirectional Encoder Representations from Transformers), and LSTM (Long Short-Term Memory) algorithms. The background of this research is the importance of understanding public opinions and reactions to sensitive and far-reaching issues, such as fuel price increases. Social media, particularly Twitter, has become a rich source of data for analyzing public sentiment, but the unstructured, dynamic and informal nature of the data requires a more comprehensive and adaptive analysis approach. This research applies its method to the context of the fuel increase results to explore public views and reactions. This hybrid approach aims to provide a omprehensive and adaptive sentiment analysis solution to overcome the challenges of dynamic Twitter data, with the results of the fuel increase as the main case study. In this Final Project, several methods are implemented to compare their accuracy results, namely Lexicon, BERT, LSTM, and Hybrid. In this Final Project, several methods were used to compare the accuracy results. For the initial method using Lexicon then for the next method BERT, then the LSTM method and finally the hybrid method. The first stage in making this Final Project is data processing, data preprocessing such as data cleansing, tokenize, stopword, model building and model testing using confusion matrix, calculating accuracy, precision, recall, and F1-score.Based on evaluation testing, the accuracy results that have the best results are in the Metode Hibrid with accuracy results using 70% training and 30% testing getting a result of 85%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	BERT, Lexicon, LSTM, Machine Learning, Twitter
Subjects:	Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) T Technology > T Technology (General) T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Aqil Ramadhan Hadiono
Date Deposited:	03 Feb 2025 07:33
Last Modified:	03 Feb 2025 07:33
URI:	http://repository.its.ac.id/id/eprint/117984

Actions (login required)

View Item