The Multimodal Dataset of Negative Affect and Aggression: A Validation Study

摘要: Within the affective computing and social signal processing communities, increasing efforts are being made in order to collect data with genuine (emotional) content. When it comes negative emotions even aggression, ethical privacy related issues prevent usage of many emotion elicitation methods, most often actors employed act out different scenarios. Moreover, for databases, emotional arousal is not explicitly checked, footage annotated by external raters based on observable behavior. In attempt gather a step closer real-life, previous work proposed an method collecting database affect aggression that involved unscripted role-plays between regulation training (actors) naive participants (students), where only short role descriptions goals given participants. this paper we present validation study investigating whether actors' behavior (e.g. becoming more aggressive) had real impact students' arousal. We found significant changes heart rate variability (HRV) parameters corresponding level states actors, therefore conclude can be considered as good candidate elicitation.