Modeling Personality Traits for Digital Humans 2007-01-2507
Present Digital Human Models (DHMs) are applicable for various purposes. Most of them are used to model anthropometry and motion behavior for workplace design and occupational health. But growing functionality has increased applications in other domains. One of these is populating Virtual Environments (VEs) or serving as advanced media for natural human-computer interaction (HCI). In this case DHMs represent real persons (e.g., local users or distant communication partners) in distributed virtual meetings. Moreover, they can be controlled by artificial intelligence in order to interact with the system and enhance HCI. The DHM turns into a Virtual Human then. Most of the Virtual Humans used for marketing nowadays have a realistic anthropomorphic appearance. In contrast to this, their behavior modeling is often restricted to general algorithmic functions with face validity only. Due to inconsistencies between appearance and behavior this might result into a loss of credibility. The integration of personality models offers a way to overcome these inconsistencies and enhance credibility and realism. Psychology provides different models to describe personality. Personality can be considered as the result of the interaction of personality attributes. There are different models for clustering attributes to traits. It has been shown that these traits strongly affect the individual behavior. By adding such individual behavior the communication between the user and the Virtual Human can be enhanced.
This paper presents the results of a review of applications and opportunities of such models for Virtual Humans. Based on these findings a novel approach for integrating personality models is introduced. The goal of the presented novel approach is to find a measurable, objective relation of single traits and motion behavior.