Sentiment analysis of foreign tourists to Bangkok using data mining through online social network

Abstract

This paper presents an analysis of sentiment of foreign tourists to Bangkok, Thailand, using data mining approach through online social networks. The objective is to acquire information on sentiment of foreign tourists in order to improve and foster tourism industry of Bangkok. This paper has retrieved 10,000 datasets from Twitter in 2017. Such datasets were tokenized and filtered in order to obtain sentiment English words. Subsequently, the sentiment English words were purposely classified into five categories of visiting Bangkok, involving (i) Traveling, (ii) Business, (iii) Visiting Family, (iv) Education, and (v) Health and Treatments. It has revealed that the traveling purpose has the highest percentage of 71.93% followed by business and visiting family. Therefore, the sentiment of foreign tourists to traveling in Bangkok was analyzed through four approaches, i.e. (i) Decision Tree, (ii) Naïve Bayes, (iii) Support Vector Machine (SVM), and (iv) Artificial Neural Network (ANN), using RapidMiner Studio7.4. The results have shown that the foreign tourists visit in Bangkok mostly for nightlife activity, Thai culture, and shopping with percentages of 65.54%, 16.07%, and 13.61%, respectively, meanwhile temple and historical sites, Thai cuisine, and nature are not significant. The accuracy of sentiment analysis approaches of Decision Tree, Naïve Bayes, SVM, and ANN are 79.83%, 55.66%, 80.11%, and 80.33%, respectively. Based upon ANN approach that provides the highest accuracy, the positive sentiments were found to be a visit for nightlife activity, temple and historical sites, Thai cuisine, and nature. On the other hand, the negative sentiment was Thai culture while shopping is relatively neutral. This paper therefore suggests an acceleration of nightlife activity of Bangkok in order to foster tourism industry of Bangkok.