KIIDS: Knowledge-Intensive Interactive Digital Storytelling

Federico Peinado and Pablo Gervás

The Knowledge-Intensive Interactive Digital Storytelling (KIIDS) model integrates automatic direction in the generic and reusable development of interactive digital storytelling applications as videogames, pieces of interactive art or educational multimedia. KIIDS consists of an extensible repository of ontologies (representing fields of knowledge about interaction, narration and simulation) and a framework of reusable software components (implementing specific techniques that are used to generate specific elements of stories for each domain and application).

One of the basic problems of interactive narrative applications is controlling the highly customized experience of the interactors at the same time as enabling the development of a good preauthored storyplot.

AI researchers have developed a number of systems that try to solve this conflict (called the Dilemma of Interactive Storytelling) in an automated way, using artificial storytellers. These systems combine a significant amount of knowledge with different reasoning algorithms to adapt multiform plots to the interactor behaviour at runtime. Each system uses its own approach, architecture, knowledge representation and implementation language, so nowadays it is virtually impossible to reuse data or code between them.

The purpouse of this model is to be a reusable software and knowledge framework for the development of these artificial storytellers, using well-known open standards.

KIIDS runs in parallel with the main process of the definitive application, establishing a dialog in which the system:

• listens to the application about the world state,
• represents internally that state using description logics,
• classifies the situation by comparing its elements with the ones in the ontology,
• decides the next steps that story must follow and
• order the application to act consequently to change the world state.

Implementations

jKIIDS is implemented in Java, using open source software (like Jena Semantic Web Framework and Pellet OWL Reasoner) to manage and reason with W3C OWL DL ontologies. Other developers can easily extend this model with more subpackages that deal with specific-domains problem-solving methods and ontologies, and reuse KIIDS as a part of their projects, thanks to its open source license.

KIIDSOnto is the OWL DL ontology of KIIDS that includes concepts, roles and a knowledge base of individuals about Interaction, Narration and Simulation for case-based and rule-based reasoning.

KIIDS is an instance of the DLApplication framework.

Applications

Interactive Narrative is a relatively new field and it is difficult to find formal studies that shows how to create this kind of stories. We have done a review over the pen-and-paper methodology of Role-Playing Games (RPGs) because Game Masters are the best models we have found in real life for designing and directing interactive stories. The aim of this project is to transfer their player modeling, their rules for interaction management and their improvising algorithms from the real world to a computer system. Now we are coding some prototypes of different applications of this system to show its functionality.

• Crossroads in Middle-Earth, automatically directed Interactive Storytelling for games
• ProtoPropp, the Fairy-Tale Generator
• Suzanne & Zu! Semiautomatic Generation of TV Cartoon's Plots

jKIIDS is under active development.
Soon you will be able to DOWNLOAD every public release of jKIIDS at Google Code

Selected publications

Mediación Inteligente entre Autores e Interactores para Sistemas de Narración Digital Interactiva (Spanish, PDF)

Transferring Game Mastering Laws to Interactive Digital Storytelling (PDF)