A Fine-tuned BERT-based Approach for Sentiment Analysis of Indonesian Public Towards ChatGPT

The rapid development of artificial intelligence technology, one of which is ChatGPT, not also invites several opinions in Indonesian society, so it is necessary to do further sentiment analysis to find out the views of the Indonesian people towards ChatGPT on popular social media for opinionated communities such as Twitter. By using the sentiment analysis method and utilizing BERT (Bidirectional Encoder Representations from Transformers) technology, this study aims to analyze the sentiments of the Indonesian people towards ChatGPT and measure the performance of IndoBERT-Base and IndoBERT-Large to obtain sentiment analysis results. Based on the experiments, the IndoBERT-Base yielded an accuracy of 78%, and IndoBERT-Large yielded an accuracy of 79% based on hyperparameter tuning.