Comparison of Methods for Classifying Comments on Tourist Places by Sentiment Analysis

Abstract

Nowadays, tourists express their experiences as opinions in various digital sources after visiting a destination, which is considered a valuable information for tourist companies or other related companies to identify which places are an opportunity for improvement, and for tourists when planning their trips. This research proposes the comparison of the support vector machine (SVM), naïve Bayes (NB), and a suggested method based on SVM and chi-square as a feature selection method. The proposed hybrid technique obtained the best result, followed by SVM and finally naïve Bayes, each with 76.50%, 67.53% and 66.91% accuracy, respectively.