Is There a Difference in the Perception of City in Pre-Pandemic and Peri-Pandemic on Social Media? Case Study from Taiwan

The purpose of this study was to consolidate machine learning applications and develop a method to simultaneously analyze unstructured text and images pertaining to travel and tourism. This paper extracted city-related tourist-generated content from social media posts and analyzed this content to elucidate public perception of Taipei and identify the factors that make these posts attractive. Amidst the global COVID-19 pandemic of the early 2020s, this study examines social media discourse on urban topics. Focused on the period from 2019 to 2020, it compares content to discern shifts in societal concerns amidst the pandemic’s progression. The analysis aims to illuminate evolving thematic patterns within city-related discussions against the backdrop of this unprecedented public health crisis. Several techniques and technologies, including content mining, Google Cloud Vision AI, topic modeling, and artificial intelligence machine learning were adopted to analyze the images and interactive characteristics of tourist-generated content relating to the city imagery and tourism transformation of Taipei. The data analyzed in this study was collected from Facebook, and RapidMiner was employed as the mining environment to apply topic modeling to identify the topics in tourist-generated content relating to Taipei before and during the pandemic and elucidate expectations and topic evolutions; and extract meaning images and text from the topics and combine them with interactive data from social media posts to identify the topics inductive to the public at different periods of the pandemic. The main graphic theme before the epidemic was to convey the charm of Taipei, compared to the graphic theme during the epidemic, which shifted to a nature-based image.