ANALISIS SENTIMEN MENGGUNAKAN SVM DAN KNN PADA REVIEW DRAMA KOREA DI MYDRAMALIST

Nur Raisa (Universitas Logistik dan Bisnis Internasional)
Noviana Riza (Universitas Logistik dan Bisnis Internasional)
Woro Isti Rahayu (Universitas Logistik dan Bisnis Internasional)

Jurnal Informatika Teknologi dan Sains (Jinteks)
Vol 5 No 4 (2023): EDISI 18

Use of K-Nearest Neighbors and Support Vector Machine methods, sentiment analysis was performed on a Korean drama review dataset found on the MyDramaList platform. This dataset contains information about Korean drama reviews provided by MyDramaList users, and is processed through text processing stages such as word beheading, stopwords removal, and cleaning. This research uses two classification methods, SVM and KNN. SVM classifies sentiment based on the feature vectors obtained, while KNN serves as a comparison to measure the performance of SVM. During experiments with test data, the performance of both methods is assessed by evaluation metrics such as accuracy, precision, recall, and f1 score. However, SVM tends to give better results compared to KNN in some cases. By combining SVM and KNN methods, this research improves sentiment analysis to analyze sentiment on Korean drama review dataset in MyDramaList.