Sentiment Analysis of Pro-Israel Product Boycott Action Using IndoBERT Method on Unbalanced Data
A boycott was an act taken to stop the purchase or use of a particular product or service as a form of public protest against a particular company or group committing a deviation. The Israeli-Palestinian conflict, which had been ongoing since 1948, peaked in October 2023 and had claimed more than 35...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | Indonesian |
| Published: |
Universitas Muhammadiyah Purwokerto
2025-08-01
|
| Series: | Jurnal Informatika |
| Subjects: | |
| Online Access: | http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/25976 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | A boycott was an act taken to stop the purchase or use of a particular product or service as a form of public protest against a particular company or group committing a deviation. The Israeli-Palestinian conflict, which had been ongoing since 1948, peaked in October 2023 and had claimed more than 35,000 Palestinian lives. This conflict generated a wide range of public opinions in Indonesia, which were expressed through social media, especially Twitter. Thus, the sentiment analysis of public reactions on Twitter became important to understand the reactions and perspectives of society towards the boycott of Pro-Israel products. This study used the IndoBERT method, which was a variant of the BERT method specifically designed to understand Indonesian. Although many studies had applied the IndoBERT method for sentiment analysis and text classification in Indonesian, none had used the IndoBERT method along with data balancing techniques to analyze Indonesian sentiments regarding the boycott of Pro-Israel products on Twitter. Therefore, this study aimed to develop a sentiment analysis model using the IndoBERT method with more data to examine sentiments related to the boycott of Pro-Israel products on Twitter using imbalanced data, as well as to evaluate the effect of balancing methods using under sampling and oversampling on the model’s accuracy and performance. The methods used included data crawling, data preprocessing, labeling with a Lexicon-Based approach, data balancing, and data splitting. The IndoBERT model was trained with 20 epochs, a batch size of 16, and a learning rate of 2e-5. The results of the study showed that the model with balanced data using the oversampling method achieved an accuracy of 97% and an F1-Score of 97%, which was better compared to the model with imbalanced data and the undersampling method. Thus, data balancing using the oversampling method proved to be effective in improving accuracy in sentiment analysis. This research made a significant contribution to understanding the behavior of Indonesian society towards a product boycott supporting Israel and suggested further exploration in parameter optimization and evaluation with larger and more diverse data, as well as further development of data balancing methods to improve the generalization and capabilities of the model |
|---|---|
| ISSN: | 2086-9398 2579-8901 |