Mining public sentiments and perspectives from geotagged social media data for appraising the post-earthquake recovery of tourism destinations

Abstract Post-disaster recovery involves interdependent processes of physical and psychological rehabilitations. Over the past few years, researchers have explored geotagged social media data to assist the planning, monitoring, and assessment of the post-disaster recovery of tourism destinations, given its advantages over traditional approaches. Nonetheless, recent studies have mostly focused on quantitatively accessing the physical elements of post-disaster recovery (e.g., infrastructure reconstruction and re-influx of tourists). Few studies have explored people's sentiments and perspectives over the process of post-disaster recovery. In this study, a mixed methods approach involving sentiment analysis and Latent Dirichlet allocation (LDA) topic modeling is designed for mining sheer volume of tweets about Lombok and Bali, generated by nonlocal Twitter users after a series of earthquakes in the two places in August 2018. The findings mainly suggest that people have generally become less negative about Lombok and Bali over time, despite fluctuations in their sentiment polarities' central tendencies. In addition, dissatisfactions about the housing reconstruction progress, tourism recovery status, and living conditions in the affected areas of Lombok still existed in 2019; contestations have been found with regard to the huge funds for hosting the 2018 Bali IMF-World Bank meeting after the earthquakes. The overall results of this study have proved that the adopted approach can effectively reveal the variations of people's sentiments and perspectives of general and specific issues regarding post-disaster tourism recovery over time.