Design Space Descriptions for Logical Generation of Content
Thomas More Smith, Julian A. Padget, Andrew Vidler · Frontiers in artificial intelligence and applications · 2016
This paper describes our research-in-progress into providing content-agnostic procedural content generation for computer games via answer set programming and a formal semantic description of the design space. Existing generative approaches are often closely coupled to a particular form of content and/or game, limiting opportunities for re-use—a dungeon level generator in one game is substantially different to a sci-fi weapon generator in another. We propose an approach involving a re-usable generator system that is able to produce and refine a model of any structured content from formal desired properties produced by designers. As an intermediate step towards this system for multi-purpose content generation we present initial work on generating content within a commercial game engine, using answer set programming to generate varied and challenging combat progressions for third-person action/combat games. By using semantic models as input for a content-agnostic generation system, we hope to provide more domain-general content generation.