<![CDATA[The implementation of the Electronic System Operator (ESO) regulation, which imposes blocking sanctions on several ESOs that do not register, has caused a variety of opinions from the public, especially on social media Twitter to raise the hashtag #BlokirKominfo. In this research, sentiment analysis was carry outed to determine the response of Twitter users to the implementation of ESO regulations by MoCI. Sentiment analysis is a textual information extraction process that classifies sentiment into positive and negative categories. The steps that are used including crawling data, text preprocessing, labeling, feature selection, term weighting with TF-IDF and classification using the Penalized Logistic Regression (PLR) with the L1 regularization and Support Vector Machine (SVM) with the RBF kernel. Sentiment classification in PLR is basically finding the optimal weight parameter. The idea of SVM sentiment classification is to find the best hyperplane to separate the data points. Evaluation of classification performance uses the accuracy value calculated through the confusion matrix. The highest percentage of accuracy in sentiment classification results using the PLR is 84,12% and SVM is 83,53%. It means that the PLR algorithm works better than the SVM algorithm in classifying public sentiment towards the implementation of ESO regulations on Twitter.]]>
Copyrights © 2023
