Yaulie Rindengan
Universitas Sam Ratulangi

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

Tourism in the Era of Industry 4.0: Harnessing Digital Storytelling and Social Media for Effective Promotion Stanss. L. H. V. Joyce Lapian; Silvya L. Mandey; Willem J. F. A. Tumbuan; Imelda W. J. Ogi; Darmawati Syam; Yaulie Rindengan; Yance Ambaho; Roberthair Suripatty; Winarsih Winarsih; Meiland Makalisang; Naomi Ormak
Public Policy : Jurnal Aplikasi Kebijakan Publik dan Bisnis Vol. 6 No. 1 (2025): Public Policy : Jurnal Aplikasi Kebijakan Publik dan Bisnis
Publisher : Univiversitas Dr. Djar Wattiheluw

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Digital technology has transformed tourism promotion, especially through digital marketing and storytelling. The COVID-19 pandemic accelerated the use of social media and e-commerce, with storytelling helping build emotional connections and improve destination images. In Indonesia, the Ministry of Tourism uses platforms like TikTok and Instagram for this purpose. Research in West Papua Barat Daya shows that short videos, photos, and blogs effectively raise awareness and attract tourists. Despite challenges like limited infrastructure and skills, digital storytelling remains a powerful and sustainable tool for tourism promotion.
Indonesian News Topic Classification Using Naïve Bayes and SVM With TF-IDF Daniel Azarya; Roynaldo Mailangkay; Beverly Rauan; Ade Yusupa; Yaulie Rindengan
Jurnal Mahasiswa Teknik Informatika Vol. 5 No. 2 (2026): Volume 5 Nomor 2 Oktober 2026
Publisher : Universitas Ngudi Waluyo

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Abstrak Pertumbuhan volume berita daring berbahasa Indonesia semakin meningkatkan tantangan dalam menyajikan konten yang relevan kepada pembaca secara tepat waktu. Sistem klasifikasi otomatis untuk berita berbahasa Indonesia yang ada saat ini masih terbatas dalam hal evaluasi komparatif yang sistematis pada beberapa kategori topik secara bersamaan. Penelitian ini menjawab kesenjangan tersebut dengan mengimplementasikan dan membandingkan dua algoritma machine learning—Naive Bayes (NB) dan Support Vector Machine (SVM)—yang dipadukan dengan ekstraksi fitur Term Frequency-Inverse Document Frequency (TF-IDF) untuk klasifikasi topik berita berbahasa Indonesia secara multi-kelas. Dataset seimbang berisi 1.000 artikel dibangun secara mandiri melalui web scraping dari Kompas.com, salah satu portal berita nasional terkemuka di Indonesia, yang mencakup lima kategori: Politik, Olahraga, Teknologi, Ekonomi, dan Hiburan (masing-masing 200 artikel). Praproses teks meliputi case folding, penghapusan noise, tokenisasi, penghapusan stopword menggunakan daftar stopword Bahasa Indonesia dari NLTK, dan stemming morfologis melalui PySastrawi. Vektorisasi TF-IDF menghasilkan matriks fitur berukuran 1.000 × 5.000 menggunakan unigram dan bigram. Ketangguhan model dievaluasi menggunakan stratified 5-fold cross-validation, dengan 20% data uji yang disisihkan untuk evaluasi akhir. SVM dengan kernel linear mencapai akurasi 94,5% (F1-score: 94,6%), mengungguli Multinomial Naive Bayes (akurasi: 89,0%, F1-score: 89,1%) pada seluruh lima kategori. Uji signifikansi McNemar mengonfirmasi bahwa selisih kinerja tersebut signifikan secara statistik (p < 0,05). Keunggulan SVM dalam memaksimalkan margin pada ruang TF-IDF berdimensi tinggi, dibandingkan asumsi independensi kondisional pada Naive Bayes, menjadi penyebab keunggulan yang teramati. Temuan ini memberikan baseline yang dapat direplikasi untuk klasifikasi berita berbahasa Indonesia dan mendukung pengembangan sistem agregasi berita cerdas yang sadar bahasa di Indonesia. Kata Kunci: Berita Indonesia; Naive Bayes; Support Vector Machine; klasifikasi teks; TF-IDF Abstract The growing volume of Indonesian online news has intensified the challenge of delivering relevant content to readers in a timely manner. Existing automatic classification systems for Indonesian-language news remain limited in systematic comparative evaluations across multiple topic categories simultaneously. This study addresses that gap by implementing and comparing two machine learning algorithms—Naive Bayes (NB) and Support Vector Machine (SVM)—combined with Term Frequency-Inverse Document Frequency (TF-IDF) feature extraction for multi-class Indonesian news topic classification. A balanced dataset of 1,000 articles was constructed independently via web scraping from Kompas.com, a leading national news portal in Indonesia, spanning five categories: Politics, Sports, Technology, Economy, and Entertainment (200 articles each). Text preprocessing included case folding, noise removal, tokenization, stopword removal using NLTK’s Indonesian stopword list, and morphological stemming via PySastrawi. TF-IDF vectorization produced a 1,000 × 5,000 feature matrix using unigrams and bigrams. Model robustness was assessed using stratified 5-fold cross-validation, with a held-out 20% test set used for final evaluation. SVM with a linear kernel achieved 94.5% accuracy (F1-score: 94.6%), outperforming Multinomial Naive Bayes (accuracy: 89.0%, F1-score: 89.1%) across all five categories. A McNemar’s significance test confirmed the performance gap is statistically significant (p < 0.05). SVM’s superior margin-maximization in high-dimensional TF-IDF space, compared to Naive Bayes’ conditional independence assumption, accounts for the observed advantage. These findings provide a reproducible baseline for Indonesian news classification and support the development of intelligent, language-aware news aggregation systems in Indonesia. Keywords: Indonesian news; Naive Bayes; Support Vector Machine; text classification; TF-IDF