The ensemble of Naïve Bayes classifiers for hotel searching

The objective of the paper is to present a new ensemble of Naïve Bayes classifiers model for an application of hotel searching. The dataset were collected from 293 reviews of 15 hotels in Phuket. The main idea of the proposed model is to combine two models of Naïve Bayes classifiers with different feature selection techniques. The output of the searching model is a list of hotel names ranking by hotel probability related to user keywords. The searching performance of the ensemble model was compared with two classical searching methods: Boolean searching and Boyer-Moore searching. The results show that the ensemble of Naïve Bayes classifiers model provides the highest average rank_accuracy. In addition, the proposed model also takes the fastest time in searching when compared with the other techniques.