Ph.D. Thesis Proposal Announcement

Title: A Computational Model of Suspense for the Augmentation of Intelligent Story Generation

Brian O'Neill
School of Interactive Computing
Georgia Institute of Technology

Date: Monday, May 7
Time: Noon--2pm
Location: TSRB 223

Committee:

• Dr. Mark Riedl (Advisor, School of Interactive Computing, Georgia Institute of Technology)
• Dr. Pablo Gervás (Departamento de Ingeniería de Software e Inteligencia Artificial, Facultad de Informática, Universidad Complutense de Madrid)
• Dr. Ashok Goel (School of Interactive Computing, Georgia Institute of Technology)
• Dr. Brian Magerko (School of Literature, Communication, and Culture, Georgia Institute of Technology)
• Dr. Ashwin Ram (School of Interactive Computing, Georgia Institute of Technology)

Abstract:
Computer scientists have tried for more than three decades to determine whether, and how, intelligent computational systems can create interesting narratives from scratch. To date, these systems have been unreliable at creating novel and aesthetically pleasing stories capable of eliciting emotional responses from human readers, in part because these systems do not adequately consider structure or audience emotion. Simply put, story generation systems do not have sufficient understanding of story aesthetics nor an understanding of how story structure affects emotional change in an audience.

Expert storytellers who craft narratives for entertainment – films, novels, games, etc. – often structure their narratives in order to elicit an emotional response from the viewer, reader or player. The idea that stories should have an emotional impact on the audience has been part of the study of drama ever since Aristotle. Authors have an array of methods for eliciting emotional responses from an audience, and one such approach is the inclusion of suspense. Despite the prevalence of suspense as a tool for storytelling, there has been little investigation of computational techniques for generating or understanding this aspect of narrative.

To address this gap, I propose a computational model of suspense based on psychological and narratological theories of suspense. Further, I propose a story generation system that makes use of this computational model as a guide in the generation process. This story generation system must be capable of understanding the structure of the story it generates, as well as understanding the effects the story may have on human emotional responses. I hypothesize that the stories generated by this system will be identified as more suspenseful than stories produced by a generation system lacking a model of suspense.