ANALISIS SENTIMEN KOMENTAR YOUTUBE TERKAIT PENERAPAN MAKAN BERGIZI GRATIS MENGGUNAKAN MODEL ALGORITMA SVM

Abstract

            Penerapan program makan bergizi gratis adalah langkah yang strategis meningkatkan kesejahteraan masyarakat, terutama kelompok rentan. Penelitian ini menganalisis persepsi sentimen masyarakat terhadap program tersebut melalui komentar YouTube menggunakan algoritma “Support Vector Machine (SVM).”Data diperoleh melalui web scraping komentar dari video YouTube yang relevan. Tahapan analisis meliputi pengumpulan data, pra-pemrosesan (pembersihan teks, penghapusan kata tidak relevan, tokenisasi), dan ekstraksi fitur menggunakan TF-IDF. SVM digunakan untuk mengklasifikasikan komentar menjadi sentimen positif, negatif, dan netral, dengan evaluasi menggunakan presisi, recall, dan F1-score. Hasil menunjukkan akurasi model sebesar 84,17%. Sebagian besar komentar bernada positif, mencerminkan dukungan publik, meski ada kritik terkait distribusi dan sosialisasi program. Komentar netral cenderung informatif tanpa opini eksplisit. Penelitian ini memberikan wawasan penting bagi evaluasi program oleh pemerintah, membantu meningkatkan kualitas implementasi melalui respon publik yang lebih terukur. Analisis sentimen berbasis algoritma terbukti efektif dalam memahami data teks berskala besar dan tidak terstruktur.

            The implementation of the free nutritious meal program is a strategic step to improve community welfare, particularly for vulnerable groups. This study analyzes public sentiment perceptions of the program through YouTube comments using the "Support Vector Machine (SVM)" algorithm. Data was collected through web scraping of relevant YouTube video comments. The analysis phases include data collection, preprocessing (text cleaning, removal of irrelevant words, tokenization), and feature extraction using TF-IDF. SVM was employed to classify comments into positive, negative, and neutral sentiments, with evaluation metrics including precision, recall, and F1-score. The results showed a model accuracy of 84.17%. Most comments were positive, reflecting public support, although there were criticisms related to program distribution and socialization. Neutral comments were generally informative without explicit opinions. This study provides valuable insights for government program evaluations, helping to improve implementation quality through more measurable public responses. Algorithm-based sentiment analysis has proven effective in understanding large-scale and unstructured text data.