Analisis sentimen komentar terhadap pilpres 2024 pada media sosial tiktok menggunakan BERT

Riyaldi, Steven (2024) Analisis sentimen komentar terhadap pilpres 2024 pada media sosial tiktok menggunakan BERT. Bachelor thesis, Universitas Pelita Harapan.

Preview	Text (Title) Title.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (15kB) | Preview
Preview	Text (Abstract) Abstract.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (446kB) | Preview
Preview	Text (ToC) ToC.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (775kB) | Preview
Preview	Text (Chapter1) Chapter1.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (755kB) | Preview
	Text (Chapter2) Chapter2.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB)
	Text (Chapter3) Chapter3.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (3MB)
	Text (Chapter4) Chapter4.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB)
	Text (Chapter5) Chapter5.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (303kB)
Preview	Text (Bibliography) Bibliography.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (489kB) | Preview
	Text (Appendices) Appendices.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB)

Abstract

Salah satu platform media sosial yang terkenal saat ini adalah Tiktok. Pada Tiktok, terdapat banyak topik yang sedang viral, salah satunya saat penelitian ini dilakukan adalah topik tentang pemilihan umum presiden dan wakil presiden atau pilpres. Oleh karena itu, banyak netizen yang memberikan komentarnya mengenai pilpres pada Tiktok, baik komentar yang positif ataupun komentar yang negatif. Komentar - komentar tersebut dapat berdampak besar dalam memengaruhi publik. Oleh karena itu, penelitian ini bertujuan untuk membuat suatu model yang dapat mengklasifikasikan komentar berbahasa Indonesia menggunakan pre-trained model IndoBERT. Dataset yang digunakan pada penelitian ini diambil dari 10 konten teratas dari #Pilpres2024 pada tiktok dengan rentang waktu dari tanggal 12 Desember 2023 hingga 24 Januari 2024. Dataset yang digunakan pada penelitian ini terdiri dari 36.991 komentar, dengan 10.600 komentar positif, 10.136 komentar netral, dan 16.255 komentar negatif. Dataset melalui tahapan preprocessing data, pelabelan data, kemudian tahapan training, validation, dan testing. Setelah itu, model kemudian di-push ke huggingface dengan nama Indonesia-Pemilu-Sentiment-Classification. Hasil pengujian model pada data testing memberikan tingkat akurasi sebesar 92,1%, presisi sebesar 92,3%, recall sebesar 92,1%, dan f1-score sebesar 92,1%. Hasil yang didapatkan dari penelitian ini membuktikan bahwa model Indonesia-Pemilu-Sentiment-Classification mampu mengklasifikasikan sentimen komentar dengan sangat akurat. / One of the famous social media platforms today is TikTok. On TikTok, there are numerous trending topics, one of which during the time of this study is the topic of the presidential and vice-presidential elections or known as Pilpres. Therefore, many netizens provide their comments about pilpres on TikTok, both positive and negative comments. These comments can have a significant impact on influencing the public. Therefore, this research aims to create a model that can classify Indonesia comments using the pre-trained IndoBERT model. Dataset used in this research were obtained from top 10 content from #Pilpres2024 in Tiktok, spanning from 12 December 2023 until 24 January 2024.The dataset used in this research consists of 36,991 comments, including 10,600 positive comments, 10,136 neutral comments, and 16,255 negative comments. The dataset goes through the stages of data preprocessing, data labeling, followed by the stages of training, validation, and testing. Afterward, the model is pushed to huggingface, named Indonesia-Pemilu-Sentiment-Classification. The results of model testing on the testing data provide an accuracy rate of 92.1%, precision of 92.3%, recall of 92.1%, and an f1-score of 92.1%. The results obtained from this research prove that the Indonesia-Pemilu-Sentiment-Classification model is able to classify comment sentiments accurately.

Actions (login required)

View Item