Analisis Sentimen Berbasis Transformer: Persepsi Publik terhadap Nusantara pada Perayaan Kemerdekaan Indonesia yang Pertama

Abstract

The inaugural Indonesian Independence Day celebration in the new capital, Nusantara, marked a historic milestone. This study analyzes public sentiment toward this event using the IndoBERT model. Data was collected from Twitter during the celebration period and classified into positive, negative, and neutral sentiments. Three main approaches were employed: IndoBERT as a baseline, IndoBERT fine-tuned with IndoNLU data, and IndoBERT applied to TextBlob-labeled data. Results indicate that the TextBlob-IndoBERT model outperforms the others, effectively processing informal Indonesian text with high accuracy. These findings provide strategic insights for the government in understanding public perception regarding the development of Nusantara and demonstrate the potential of Transformer-based sentiment analysis for the Indonesian language. The study recommends further exploration of factors influencing sentiment and analysis on other social media platforms.