Analysis of Tourists' Nationality Effects on Behavior-based Emotion and Satisfaction Estimation

Smart tourism is attracting attention of researchers in recent years. Its technologies can be used by tourists in order to obtain useful information during sightseeing with smart devices etc. To provide suitable and personalized tourism information according to the situation of tourists, understanding psychological status during sightseeing, especially, emotional status and satisfaction level, is important. We assume that the psychological status of tourists is appearing and represented through unconscious behaviors during sightseeing such as head/body movements and facial/vocal expressions, and have proposed methods to estimate emotion and satisfaction statuses by sensing and analyzing tourists' behaviors. Through in-the-wild experiments with 22 participants, we found that the difference in tourists' attributes might give effects for the estimation. In this paper, we have statistically analyzed those effects, focusing on tourists' nationality. As a result of the two-way ANOVA, we found the interaction effect (disordinal interaction) between tourists' nationality and estimation performance, the main effect in differences of features, and the main effect in differences of tourists' nationality. The results imply that we need to take tourists' nationality into account for building estimation models. Contribution: We have statistically analyzed the nationality effects on tourist emotion and satisfaction estimation, and confirmed significant differences in feature contributions for estimation models.