<![CDATA[Social media has become the primary arena for the public to express opinions on government policies. This study aims to analyze public sentiment toward government policies using the Long Short-Term Memory (LSTM) model, while also examining the role of language in shaping public opinion. Data were collected from social media posts related to economic, social, and health policies, followed by preprocessing stages including text cleaning, tokenization, stopword removal, and word embedding with Word2Vec. The LSTM model was compared with Support Vector Machine (SVM) and Naïve Bayes to evaluate accuracy and performance. The results indicate that public opinion is dominated by negative sentiment (45%), particularly regarding economic policies. The LSTM model outperformed the benchmarks with an accuracy of 86.9%, surpassing SVM and Naïve Bayes. Linguistic analysis revealed the frequent use of emotional diction, sarcasm, and economic burden narratives that reinforced public resistance, while colloquial language was found to be an effective tool for engaging younger generations. This study contributes to the advancement of sentiment analysis in the Indonesian language using deep learning and provides practical recommendations for policymakers to design more persuasive and participatory communication strategies.]]>
Copyrights © 2026
