<![CDATA[Mosquito-borne diseases, including malaria, dengue hemorrhagic fever (DHF), and chikungunya, remain significant public health challenges in Indonesia. With the extensive use of the internet, social media platforms such as X (Twitter) and Instagram have emerged as vital sources for capturing public sentiment in real-time. This study aims to identify and analyze Indonesian public sentiment regarding these diseases using a Deep Learning approach. The dataset comprises 1,800 records collected between 2024 and 2025 via web scraping. The methodology utilizes the BERT (Bidirectional Encoder Representations from Transformers) model for contextual feature extraction, evaluated through three Recurrent Neural Network (RNN) architectures: LSTM, GRU, and BiLSTM. The results reveal a predominant negative sentiment (56.99%), followed by neutral (32.99%) and positive (10.01%). While all models achieved high training accuracy (approx. 96%), the Gated Recurrent Unit (GRU) demonstrated superior generalization with the highest average validation accuracy of 94.24%. This study concludes that the GRU model is the most stable and effective architecture for classifying Indonesian public health sentiments on social media.]]>
Copyrights © 2026
