Mining Tourist's Perception toward Indonesia Tourism Destination Using Sentiment Analysis and Topic Modelling

Abstract

Indonesia's Tourism industries is the second-best contributor to country's foreign exchange income for years. The growth it produced were in range of 7 to 10 percent per year since 2009. In reality, Bali is always at the top of the mind among majority of international tourists inspite of many hidden gems with spectacular qualities to rival Bali's popularity. Therefore, Indonesia's government has set new tourist destinations to increase their presence thus increasing visitation numbers. Researches on utilizing big data to support industry 4.0 model in tourism businesses are encouraged as part of national research priorities. Data Analytics models such as Sentiment Analysis and Topic Modelling can be used to reveal hidden patterns from abundant user-generated content data available in social media sites, one of which, TripAdvisor. This study aims to mine the visitors' perceptions of 10 most visited sites in Indonesia. Emotions and topics discussed in comments are two features to be extracted. Using data mining framework, five types of emotion and topics related to tourism were discovered. Data collection was done using Parsehub to acquire 3494 comments generated in 4 years. Both Sentiment Analysis and Topic Modelling were processed in Orange 3. Results shows that Joy is the most prominent emotion accompanying visitors' experiences. Topic modeling shows several important keywords toward preferences. Results of this study can be used to improve Indonesia's tourism stakeholder's decision quality especially in terms of marketing and operations.