Qualifying and Quantifying Interestingness in Dramatic Situations

Dramatic situations have long been studied in Drama Studies since they characterize tension and interestingness in a plot. In the field of Interactive Digital Storytelling (IDS), integrating knowledge about dramatic situations is of great relevance in order to design improved systems that dynamically generate more narratively-relevant events. However, current approaches to dramatic situations are descriptive and not directly applicable to the field of IDS. We introduce a computational model that fills that gap by both describing dramatic situations visually and providing a quantitative measure for the interestingness of a plot. Using a corpus of 20 Aesop's fables, we compared the calculations resulting of the model with the assessments provided by 101 participants. Results suggest that our model works appropriately at least for stories characterized by a strong plot structure rather than their semantic content