Finding tourism niche on image-based social media: Integrating computational methods

Abstract

The purpose of this research is to examine one of the most effective approaches for locating niche tourism attractions that varies by people, using a methodology that combines statistical analysis, deep learning visual image detection, and text mining. Using 30,013 posts with the hashtag #Seoul in English, the analysis focused on the Instagram posts’ time, dominant color, image visual content, and hashtag to identify niche tourism attractions. The analysis result shows that Instagram posts hashtag #Seoul that depicted “young women” and was uploaded in the evening with warm colors such as orange, yellow, and green received more “likes” than other postings. Furthermore, deep learning and text mining analysis were used to identify and forecast the actual image with the most likes in each sectoral domain, as classified by topic modeling, such as “young, woman, outdoor” and “table, plate, indoor.” Through these findings, this study identified niche hotspots of tourism attractions based on those destination image attributes in Instagram photos, which contributes to the popularity of Instagram postings. The methods and results will be particularly useful to marketers and researchers looking to uncover specialized tourism themes and combine popularity measurement with visual image analysis.