Fine-Grained Sentiment Classification of Public Opinion on Electric Cars in Indonesia Using IndoBERT

This research aims to analyze public sentiment towards electric cars in Indonesia using the IndoBERT model. Data were collected from social media platforms, especially Twitter or X, and processed using the IndoBERT Model for Indonesian-language public opinion. The dataset consists of 1,087 tweets, which were preprocessed and labelled into five sentiment categories: very positive, positive, neutral, negative, and very negative. According to the study's findings, the majority of public opinion in Indonesia is neutral (70.45%), then positive (21.65%), negative (6.13%), very positive (1.49%), and very negative (0.28%). Although the overall trend is positive, there are concerns about the charging infrastructure and the price of electric cars. The enhanced IndoBERT model achieved outstanding performance, with accuracy increasing from 74% in the first epoch to 80% in the fifth epoch and F1 score increasing from 0.83 to 0.89. These findings provide valuable insights for policymakers and manufacturers in supporting the adoption of electric cars in Indonesia