Towards a Model-Learning Approach to Interactive Narrative Intelligence for Opportunistic Storytelling

Abstract

Opportunistic storytelling is an approach to interactive narrative where game play is the ordinary activity that underlies notable story events, and the AI challenge is to tell a story about what the player is doing, that meets authorial goals. In this preliminary work, we describe a game and AI system that motivates the need for event prediction within the game world, and provides the opportunity for automated machine learning of such a predictive model. We report results showing how different feature models can be learned and compared in this context, towards automating model selection.