Text-Based Emotion Sentiment Analysis Pada Komentar Youtube Dengan Pendekatan Deep Learning: Akuisisi Tiktok Terhadap Tokopedia

Pornama, Angga (2024) Text-Based Emotion Sentiment Analysis Pada Komentar Youtube Dengan Pendekatan Deep Learning: Akuisisi Tiktok Terhadap Tokopedia. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 20082010125.-cover.pdf Download (2MB)
	Text (Bab 1) 20082010125.-bab1.pdf Download (393kB)
	Text (Bab 2) 20082010125.-bab2.pdf Restricted to Repository staff only until 15 December 2026. Download (935kB) \| Request a copy
	Text (Bab 3) 20082010125.-bab3.pdf Restricted to Repository staff only until 15 December 2026. Download (1MB) \| Request a copy
	Text (Bab 4) 20082010125.-bab4.pdf Restricted to Repository staff only until 15 December 2026. Download (4MB) \| Request a copy
	Text (Bab 5) 20082010125.-bab5.pdf Download (438kB)
	Text (Daftar pustaka) 20082010125.-daftarpustaka.pdf Download (404kB)
	Text (Lampiran) 20082010125.-lampiran.pdf Restricted to Registered users only until 15 December 2026. Download (525kB) \| Request a copy

Abstract

The acquisition of a 75% stake in Tokopedia's by TikTok has sparked various public responses, both pros and cons, with concerns related to potential monopolies, unfair trade competition, data security, and China's geopolitics. This study aims to analyze public sentiment on the issue using an emotion-based approach. Data consisting of 3,041 YouTube comments were collected through scraping and processed using deep learning algorithms, namely CNN, LSTM, CNN-LSTM, and LSTM-CNN, with TF-IDF feature extraction and data augmentation. It revealed that the majority of emotions in the dataset were negative, such as apprehension, disapprove, anger, and fear, reflecting public disapproval of the acquisition. The main concern revolves around TikTok’s full control over Tokopedia, which could potentially harm the local platform and business actors within the e-commerce ecosystem. The scenario of converting emojis into text in the dataset tended to degrade model performance due to increased noise and loss of emotional meaning. The CNN model demonstrated the best performance, achieving the highest accuracy of 97.03% with x3 augmentation, F1-Score of 0.97, precision of 0.97, and recall of 0.96, indicating that CNN outperformed other models in identifying emotion classes in the data.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Agussalim, Agussalim	NIDN0911088501	agussalim.si@upnjatim.ac.id
UNSPECIFIED	Wati, Seftin Fitri Ana	NIDN0020039104	seftin.fitri.si@upnjatim.ac.id

Subjects:

Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Q Science > QA Mathematics > QA76.87 Neural computers
T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Departemen of Information Systems

Depositing User:

Angga Pornama

Date Deposited:

16 Dec 2024 05:46

Last Modified:

16 Dec 2024 05:46

URI:

https://repository.upnjatim.ac.id/id/eprint/33433