Sentiment Analysis and Topic Modeling of Public Opinion on Indonesia New Capital City Development Policies

Abstrakt

This study investigates public sentiment dynamics and dominant thematic concerns related to Indonesia’s new capital city development project (Ibu Kota Nusantara–IKN), particularly in the context of the political leadership transition from President Joko Widodo to President-elect Prabowo Subianto. Utilizing a dataset comprising 9,451 tweets collected from 2017 to 2025, sentiment analysis and topic modeling were applied to classify sentiment polarity and identify prevailing public discourse themes. Various traditional machine learning models—including Naïve Bayes, Support Vector Machine (SVM), AdaBoost, XGBoost, and LightGBM—were systematically compared with transformer-based deep learning models, specifically IndoBERT, to determine their effectiveness in sentiment classification. Results demonstrated that the IndoBERT model outperformed all traditional classifiers, achieving the highest accuracy, precision, recall, and F1 score, highlighting its superior capability in capturing nuanced linguistic patterns within informal social media texts. Independent samples t-tests revealed statistically significant sentiment shifts between the two political phases, emphasizing the impact of leadership transitions on public sentiment. Topic modeling further identified critical themes such as environmental sustainability, socio-economic implications, transparency, governance, and infrastructure development as central concerns driving public discussions. These findings provide actionable insights for policymakers and stakeholders, underscoring the importance of strategic communication and responsiveness to public sentiment in large-scale government initiatives.