<![CDATA[Rupiah redenomination has resurfaced as a strategic issue in Indonesian monetary policy following the announcement of the roadmap by Minister of Finance Purbaya Yudhi Sadewa in 2025. This study aims to analyze public sentiment towards the redenomination issue through a hybrid approach: lexicon based (LBA) based on InSet Lexicon enriched with 217 contextual phrases (“bunker money”, “DPR ketar-ketir”), combined with Naive Bayes based on TF-IDF representation. The primary dataset of 1,087 public comments from YouTube (Isu_Ekonomi_Redominasi.csv) was processed using Sastrawi (stopword removal and stemming). The results show a dominance of positive sentiment (58.3%), driven by the justice frame narrative (redenomination as a tool for exposing corrupt assets), while negative (24.1%) and neutral (17.6%) sentiment reflect concerns over inflation risks and transition confusion. The MNB model achieved an average accuracy of 86.3% in 10-fold cross-validation. The findings reveal that public support is not merely monetary, but rather an expression of collective aspirations for state transparency and accountability. This research demonstrates the effectiveness of a hybrid lexicon-enhanced approach for domain-specific sentiment analysis in Indonesian, while also providing evidence-based policy insights for inclusive policy design.Keywords: Public Sentiment; Twitter Media; Naive Bayes; VisualizationAbstrakRedenominasi rupiah kembali mencuat sebagai isu strategis dalam kebijakan moneter Indonesia pasca pengumuman roadmap oleh Menteri Keuangan Purbaya Yudhi Sadewa pada 2025. Penelitian ini bertujuan menganalisis sentimen masyarakat terhadap isu redenominasi melalui pendekatan lexicon based berbasis InSet Lexicon yang diperkaya dengan 217 frasa kontekstual (misal: “uang bunker”, “DPR ketar-ketir”), dikombinasikan dengan Naive Bayes berbasis representasi TF-IDF. Dataset primer berupa 1.087 komentar publik dari YouTube (Isu_Ekonomi_Redominasi.csv) diproses menggunakan Sastrawi (stopword removal dan stemming). Hasil menunjukkan dominasi sentimen positif (58,3%), didorong oleh narasi justice frame (redenominasi sebagai alat eksposur aset korupsi), sementara sentimen negatif (24,1%) dan netral (17,6%) mencerminkan kekhawatiran atas risiko inflasi dan kebingungan transisi. Model Naive Bayes mencapai akurasi rata-rata 86,3% dalam 10 fold cross validation. Temuan mengungkap bahwa dukungan publik tidak hanya bersifat teknis moneter, melainkan ekspresi aspirasi kolektif terhadap transparansi dan akuntabilitas negara. Penelitian ini membuktikan bahwa pendekatan lexicon efektif untuk analisis sentimen domain spesifik dalam Bahasa Indonesia, sekaligus memberikan policy insight berbasis bukti untuk desain kebijakan inklusif. ]]>
Copyrights © 2026
