Perso2U: Exploration of user emotional states to drive interface adaptation

Taking into account dynamic user properties such as emotions for interfaces adaptation at runtime is a challenging task. To deal with this issue, we propose to personalize user interfaces at runtime based on user’s emotions. This approach depends on emotion recognition tools to allow an Inferring Engine to deduce user emotions during the interaction. However, this inference releases many emotions without aggregating them. It makes more difficult the interpretation of user experience. Thus, we explore the feasibility of inferring similar emotional states (negative, positive and neutral) by grouping individual emotions. To achieve our goal, this paper reports on the results of an experiment to compare detected emotional states from different face recognition tools in web interaction. It evidences that it is feasible to infer similar emotion states (positive, negative, and neutral) from different emotion recognition tools, and the level of this similarity is still premature to have a robust categorization.