Towards Generating Novel Games Using Conceptual Blending

We sketch the process of creating a novel video game by blending two video games specified in the Video Game Description Language (VGDL), following the COINVENT computational model of conceptual blending. We highlight the choices that need to be made in this process, and discuss the prospects for a computational game designer based on blending. Introduction This paper outlines the process of creating a novel video game by blending two video game descriptions given in the Video Game Description Language (VGDL). Several issues need to be considered: which games to blend? Which elements of the game will be blended and which removed? Do we need to adjust the resulting game in order to make it playable? Taking two games from the GVG-AI Competition (Perez et al. 2015), we provide a worked example based on a recent computational, domain-independant model of conceptual blending (Bou et al. 2015). Our account provides a high-level sketch of how a computational system could blend game designs, and does not propose a concrete architecture. Instead, our intent is explore how a game blending system could work. We highlight the specific challenges presented by various steps in the blending process, and describe how certain steps provide the opportunity to add or close off new potential mechanics and meanings. The implementation and evaluation of this approach will be the focus of future work. Background Recent research (Treanor et al. 2012; Cook and Colton 2014) has examined computational generation of games from provided concepts. The current exploration of blending game descriptions is motivated by our observation that game designers often build new games by combining ideas from existing games. Conceptual blending is the notion that new concepts can be generated by combining elements of existing concepts. COINVENT is an EU FP7 project that aims to develop a “computationally feasible, formal model of conceptual Copyright c © 2015, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. blending” (Schorlemmer et al. 2014)1. The project takes Goguen’s formalisation of conceptual blending as a starting point (Goguen 1999). Early implementations draw on the related notion of amalgams (Ontañón and Plaza 2010). The core ideas in the model are: Conceptual spaces may be mathematical theories, ontologies, or, in our case, video game descriptions. Signature morphisms between conceptual spaces are mappings from the symbols of a source conceptual space into the symbols of target conceptual space. In the diagram below, φ1, φ2, ψ1, ψ2, and ι are all signature morphisms. Generic spaces are a particular kind of conceptual space that possess maximum commonality with a number of given input spaces. In the diagram below, φ1 and φ2 represent the inclusion of the elements of the generic space G in both X and Y . Blends are unique conceptual spaces that preserve structure from the input spaces and a generic spaceG, via a specific arrangement of signature morphisms; namely, such that the diagram is commutative, and that ‘axioms’ in the input spaces X,Y are sent to ‘theorems’ in the unique colimit B, known as the blend: