Evaluation of a virtual Narrator's expressiveness in terms of suspense signaling

In this paper, we describe the results of the evaluation of a Virtual Character, used as a human-like, automated narrator with focus on the signaling of suspenseful situations within stories to the Virtual Narrator's audience. Signaling is done via facial expression and gestures, driven by an underlying story examination mechanism. This mechanism uses an automatic generated story to get points of suspense progression, conflicts between story actors, and the relevance of story fragments. It indicates these points to an expressiveness module, which assigns expressiveness states to the Virtual Narrator.