A Computational Model of Narrative Generation for Suspense

Although suspense contributes significantly to the enjoyment of a narrative by its readers, its role in dynamic story generation systems has been largely ignored. This paper presents Suspenser, a computational model of narrative generation that takes as input a given story world and constructs a narrative structure intended to evoke the desirable level of suspense from the reader at a specific moment in the story. Our system is based on the concepts that a) the reader’s suspense level is affected by the number of solutions available to the problems faced by a narrative’s protagonists, and b) story structure can influence the reader’s narrative comprehension process. We use the Longbow planning algorithm to approximate the reader’s planning-related reasoning in order to estimate the number of anticipated solutions that the reader builds at a specific point in the story. This paper discusses our preliminary results and concludes with suggestions for further study.