Applying Text Mining Techniques for Sentiment Analysis of Museum Visitor Reviews

In an era where information is obtained through multiple channels, more tourists gather experiential travel information from travel evaluation websites (such as TripAdvisor and Google Maps) in addition to traditional tourist attraction official websites. The information is related to the operation and management of scenic spots. Text mining techniques perform well in analyzing unstructured data so are a feasible method to analyze such review data. Therefore, we analyzed the review data using text mining to explore word segmentation, TF-IDF vectors, feature selection, keyword co-occurrence, topic modeling, sentiment scoring models, regression analysis, and association rules. We calculated the sentiment scores of tourists for the attractions and explored the relationships among the features of review content based on tourists' reviews about the attractions. With the results, we generated review topic maps and examined the relationships between features and sentiment scores. Tourist reviews of eight world-renowned museums were collected from TripAdvisor. The data set was composed of approximately 200,000 review reviews.