Janet Murray


The Schema for Narrative Authoring and Presentation Systems (SNAPS) Project is aimed at making the tools that will allow the conventions of narrative and games to be of use for the purposes of knowledge creation.  It is based on the assumption that narrative structures are abstraction systems and that narrative abstraction has been adapted to multiple media through specific technologies, such as the hexameters of epic poetry, the  first person narration of confessional novels, the over the shoulder shot of classic film. This project will explore techniques for advancing the development of similar technologies of expression for the computational medium by establishing schema and design principles for authoring and presentation environments that augment our ability to author more complex procedurally-based stories, and to navigate multiple versions of the same event and multiple points of view in such a way that we build up more complex models of the world.

This project is aimed at creating Design Schemas and Design Principles for the Creation and Transmission of Knowledge through Parameterized Narratives.  The results of this investigation will be design schemas in the form of  (1) prototypes for authoring environments that will put the procedural, simulation-building power of the computer in the hands of domain experts and writers allowing them to create multisequential and multiform scenarios without requiring them to program; (2) prototypes of presentation environments for users of such scenarios that will foreground the contrasts between instantiations, providing juxtapositions that will clarify complex causal chains and interrelationships in multi-causal scenarios;  as well as (3) a summary set of design principles illustrated by appropriate schematic mockups that will contextualize these prototypes and focus a larger conversation on how to use narrative abstraction and story-related game mechanics to support the understanding of complex systems.

SNAPS is intended as a larger research effort, but this exploratory phase will test the viability of this approach by drawing on and integrating work done by the project director over 25 years for a variety of purposes, work done by others in creating the narrative computational systems,  and on going as well as new work in narrative systems in the Georgia Tech Graduate Program in Digital Media.  The SGER project will survey and evaluate these existing examples for generalizable approaches in the form of schemas, cognitive frameworks that are also specific media-appropriate design strategies for authors and consumers of multiform (affording different instantiations of objects, characters, and events)  and multisequential (affording multiple coherent paths for traversing the same events) narratives.  For example, an earlier project, Character Maker supports the creation of Eliza-like conversations created as a set of topics and key words, with many ways to traverse the network of topics. To allow an author to see all the paths that were possible, and to add or subtract paths appropriately, the developers implemented an overview feature that displays the directed graph of all the paths connecting any two topics.  This project will explore schemas for a similar system that could be based on plot events designed as pre- and post- conditional nodes. This would allow for more coherent unfolding of plot sequences than has been achieved by previous AI-based story-plotters, because the narrative abstraction will be overseen by human intelligence, leaving the AI to do more focused, supportive tasks such as generating the larger network of possibilities and determining how they should be presented for evaluation.  Such a division of labor will also be productive for computer science, presenting AI research with well-formed problems rather than vaguer expectations of creativity.


NFS   Digital Media