Exploring destination image of dark tourism via analyzing user generated photos: A deep learning approach

Abstract

Tourism destination image (TDI) in the dark tourism context is considered to be a complex and controversial, yet rarely studied issue in the literature. This study selected three types of dark tourism destinations in China to explore TDI through analyzing user generated photos via DeepSentiBank, a method based on deep convolutional neural networks. Based on a content analysis, this study identified 11 categories of cognitive images, and found that memorial space & sculptures, commemorative symbols, and historical events & place functions were the distinctive categories of cognitive images. Based on a sentiment analysis, it revealed 24 emotions of affective images, and found that negative emotions weighed more heavily than positive emotions in dark tourism destinations. Complex network analysis further revealed multiple inter-linked relationships between cognitive and affective attributes. This study contributes to photo-based sentiment analysis in tourism research, and the findings provide insights for TDI development and management for dark tourism destinations.