Sentiment Analysis of Indonesian Movie Trailer on YouTube Using Delta TF-IDF and SVM
Muhammad Alkaff, Andreyan Rizky Baskara, Yohanes Hendro Wicaksono
Abstract
YouTube is one of the most effective social media sites for promoting products, one of which is movies. The film industry usually publishes video trailers on YouTube to promote their upcoming film. The comments that appear on YouTube could help movie producers to estimate how the public will react to their movie once it is released. In this study, we conducted a sentiment analysis on the comments of Indonesian movie trailers on YouTube. We split movie comments into four popular movie genres: action, romance, comedy, and horror. Then, we use the Delta TF-IDF word weighting method and combine it with several classification methods to compare the model performance. Finally, we evaluated the model using Stratified K-Fold cross-validation with K = 10. Results showed that Logistic Regression and Naïve Bayes are better when classifying sentiment for a specific genre. Simultaneously, the SVM model gives good performance on sentiment analysis for a more general genre.