Analisis Sentimen Masyarakat Terhadap Fenomena Perubahan Iklim Menggunakan Bert-Cnn

Abstract

Climate change is one of the most talked about global issues on social media, such as X. Despite the scientific consensus that climate change is caused by human activities, public opinion remains divided, dan such differences not only affect individual opinions, but can also have an impact on environmental policy-making. To understand public opinion, various NLP models are developed, but the challenge in using traditional NLP models is their limitation in capturing contextual information in a sentence. Therefore, a BERT-CNN hybrid model is proposed with the assumption that the contextual embedding in BERT along with the feature extraction capability of CNN is able to classify public sentiment on the issue of climate change. The method used starts with data preparation, model design, model training, model testing, dan comparison of the main model with other models. The experimental results show that the BERT-CNN model is quite good at classifying public sentiment related to climate change, as evidenced by its f1-score value of 0,7197. The model’s performance on minority classes is better than BERT or CNN model on it’s own. Thus, this research provides new insights into the development of deep learning-based sentiment analysis methods on the issue of climate change.