Sentiment Rating Analysis on Videos on Youtube Social Media Using STRUCT-SVM

Abstract

Sentiment analysis on comments can be used to determine sentiment rating. The comments used are comments on Youtube. The type of video used is the official trailer video Indonesian movie. This paper contains steps to determine sentiment rating by notice the structure of comments. The structure of comments is needed because not all comments are relevant to the topic. Classes on comments are divided into seven classes including positive films, neutral films, negative films, positive not films, neutral not film, negative not film, and spam / off-topic. Comments that have a positive film or film negative class are used to determine sentiment rating. The number of likes in comments also determines the sentiment rating. Comment classification using STRUCT-SVM. The results of STRUCT-SVM show accuracy of 69.68% for linear kernels and 71.34% for RBF kernels.