Multi-Aspect Sentiment Analysis Hotel Review Using RF, SVM, and Naïve Bayes based Hybrid Classifier

In the hotel tourism sector, of course, it cannot be separated from the role of social media because tourists tend to share experiences about services and products offered by a hotel, such as adding pictures, reviews, and ratings which will be helpful as references for other tourists, for example on the media online TripAdvisor. However, tourists' many experiences regarding a hotel make some people feel confused in determining the right hotel to visit. Therefore, in this study, an aspect-based analysis of reviews on hotels is carried out, which will make it easier for tourists to determine the right hotel based on the best category aspects. The dataset used is the TripAdvisor Hotel Reviews dataset which is already on the Kaggle website. And has five aspects, namely Room, Location, Cleanliness, Registration, and Service. A review analysis was carried out into positive and negative categories using the Random Forest, Support Vector Machine, and Naive Bayes Hybrid Classifier-based methods to solve this problem. In this study the Hybrid Classifier method gets better accuracy than the classification using one algorithm on multi-aspect data, namely the Hybrid Classifier gets an average accuracy of 84%, Naive Bayes gets an average accuracy of 82.4%, Random Forest gets an average accuracy of 82.2%, and Support Vector Machine get an average accuracy of 81%.