DETEKSI SARKASME BAHASA INDONESIA PADA ULASAN PRODUK MENGGUNAKAN NLP DAN SVM
Jurnal Informatika, Sistem Informasi dan Kehutanan (FORSINTA) · 30.10.2026 · 5 (2) · 85-93
DOI: 10.53978/jfsa.v5i2.836 ↗Kurz gefasst
Fachartikel in der Zeitschrift Jurnal Informatika, Sistem Informasi dan Kehutanan (FORSINTA), online erschienen am 30.10.2026. Verfasst von Dea Tri Erisma und 1 weiteren Autor:innen.
Abstract (Originalsprache)
The growth of e-commerce and digital platforms has increased the amount of Indonesian-language product review text available for analysis. Product reviews may contain informal expressions and sarcasm, in which the literal meaning differs from the intended meaning, making conventional sentiment analysis less reliable. This study develops a sarcasm detection system for Indonesian product reviews using Natural Language Processing (NLP) and Support Vector Machine (SVM). The research uses public datasets, including Tokopedia product reviews and the Indonesian Dataset for Sarcasm Detection, with 139,770 labeled records. Text preprocessing consists of case folding, cleansing, stopword removal, stemming, and normalization, while tokenization is handled during TF-IDF vectorization. The resulting features are classified using a Linear Support Vector Machine with balanced class weights and parameter optimization through GridSearchCV and stratified cross-validation. The reported evaluation obtained 97.8% accuracy, with 86.5% precision, 88.8% recall, and 87.6% F1-score for the sarcasm class. A separate test using 13 manually selected cases produced 7 correct predictions (53.8%), indicating that newly encountered ironic, hyperbolic, and contradictory expressions remain challenging. The model was also implemented in a Streamlit web interface to provide real-time predictions and confidence information.
Zitieren
APA
Erisma, D. T., & Zulherry, A. (2026). DETEKSI SARKASME BAHASA INDONESIA PADA ULASAN PRODUK MENGGUNAKAN NLP DAN SVM. Jurnal Informatika, Sistem Informasi dan Kehutanan (FORSINTA), 5(2), 85-93. https://doi.org/10.53978/jfsa.v5i2.836
Vancouver
Erisma DT, Zulherry A. DETEKSI SARKASME BAHASA INDONESIA PADA ULASAN PRODUK MENGGUNAKAN NLP DAN SVM. Jurnal Informatika, Sistem Informasi dan Kehutanan (FORSINTA). 2026;5(2):85-93. doi:10.53978/jfsa.v5i2.836
BibTeX
@article{Erisma2026,
author = {Erisma, Dea Tri and Zulherry, Andy},
title = {DETEKSI SARKASME BAHASA INDONESIA PADA ULASAN PRODUK MENGGUNAKAN NLP DAN SVM},
journal = {Jurnal Informatika, Sistem Informasi dan Kehutanan (FORSINTA)},
year = {2026},
volume = {5},
number = {2},
pages = {85--93},
doi = {10.53978/jfsa.v5i2.836}
}
EUVIMED