Tourist Satisfaction Research Based On Data Mined From Web Social Platforms

Tourist satisfaction has always been a hot issue in tourism research. Whether domestically or abroad, there have been plenty of valuable researches. Mature tourist satisfaction theory system was formed, and established solid theoretic foundation. However, related researches were mostly done without quantitative methods based on massive data. This is because of two difficulties: first, the limitation of questionnaires being the sole evaluation method, second, the collection of data. This paper presented a method to approach tourist satisfaction analysis using state-of-the-art techniques in computer science and natural language processing(data-mining and sentiment analysis), combined with content analysis methodology, to efficiently gather data, and to objectively analyze them, a method in which credibility and adaptability can be ensured. The result of this page has unique value among researches in tourists who are going abroad. Especially, the successful use of computer technologies such as sentiment analysis has broadened the perspective of satisfaction analysis applications.The target of this research is Chinese tourists in six destinations(namely Hong Kong, Macau, Korea, Taiwan, Thailand, and Japan). Using a computer program, a total of 828,113 entries of ’Weibo’ are gathered from Sina Weibo Places. Using Maximum Entropy Modeling(Machine Learning in Computer Science), 12,000 random entries were chosen for manual tagging and automated language feature extraction. They serve as training and evaluation data, to produce a classification model that is capable of sentiment analysis on such data. Based on this model, all entries were automatically classified as positive, neutral, or negative. The results of classification served as the input of a "Cost versus Gain" model, which eliminated the factor of consumer pricing and was capable of comparing tourist satisfaction between different destinations(or comparing price-performance ratio in terms of tourism). Eventually the model output provides suggestions to maximize satisfaction.Based on calculated results, the ranking of tourist satisfaction from high to low is: Thailand, Taiwan, Korea, Macau, Japan, and Hong Kong. On price-performance ratio comparison, due to the fact that tourists in Hong Kong and Macau are mostly same-day visitors, their standards of expense statistics are drastically different from the rest of destinations. They are only compared with each other. Macau is superior to Hong Kong in terms of price-performance. The ranking of the rest is: Thailand, Taiwan, Japan, and Korea.From observed difference between pure satisfaction ranking and price-performance, the factor of expense does have major impact on satisfaction. The "Cost versus Gain" model can be used to horizontally compare satisfaction among different destinations in a fair way.The conclusion proved that sentiment analysis has great advantage when combined with satisfaction research. It greatly improves survey efficiency while provides more meaningful results. The application of this technique provides extra value when dealing with difficulties in tourism theory research.