ASP Narrative Planners for Measuring Solution Density

• Cory Siler
• Stephen G. Ware

Narrative generation algorithms fall on a spectrum from strong autonomy, where each character is implemented as a fully autonomous agent, to strong story, where a central agent guides all character behavior. This project compares four different approaches to narrative generation: multi-agent systems, drama managers, narrative planners, and classical planners. Each of these approaches has a different definition of what counts as a solution to the story generation problem. This project developed Answer Set Programs to generated all possible solutions for each of the above systems and compare them. We are interested to know, for example, how often a classical planner generates stories that also could have been generated by a multi-agent system, and vice versa.