Goal-Driven Conceptual Blending: A Computational Approach for Creativity

Conceptual blending has been proposed as a creative cognitive process, but most theories focus on the analysis of existing blends rather than mechanisms for the efficient construction of novel blends. While conceptual blending is a powerful model for creativity, there are many challenges related to the computational application of blending. Inspired by recent theoretical research, we argue that contexts and context-induced goals provide insights into algorithm design for creative systems using conceptual blending. We present two case studies of creative systems that use goals and contexts to efficiently produce novel, creative artifacts in the domains of story generation and virtual characters engaged in pretend play respectively.