Analisis Sentimen Ulasan Traveloka Menggunakan Metode Naïve Bayes Classifier dan Information Gain

Abstract

In the increasingly rapid digital era, Traveloka is present as an online travel agency that makes it easier for users to order and plan trips. Reviews left by users can reflect the user's experience in using the platform. Indirectly, reviews can also reflect user satisfaction. Therefore, it is important to carry out sentiment analysis of existing reviews so that you can improve service quality. This research examines the performance of the Information Gain feature selection in classifying the sentiment of Traveloka application reviews using the Naïve Bayes method. The research results show that classification using the Naïve Bayes model obtained an accuracy of 83%, precision of 81%, and recall of 98%. Meanwhile, classification with feature selection obtained an accuracy of 79%, precision of 76%, and recall of 100%. This shows that the feature selection performance has not been able to increase the accuracy value.

Keywords: Sentiment Analysis, Reviews, Traveloka, Naïve Bayes Classifier, TF-IDF, Information Gain