A hybrid approach to model a bayesian network of culture-specific behavior

Cultural background can crucially influence human behavior. For virtual characters, that are supposed to imitate human behavior in a natural manner, integrating cultural background came into focus the recent years. Unlike earlier approaches, this paper presents a hybrid approach of modeling a Bayesian network that generates culture-specific behaviors for virtual characters. For the hybrid approach, the structure of the network is designed based on social science theories, while its parameters are derived from a video corpus.