BioSpark: An End-to-End Generative System for Biological-Analogical Inspirations and Ideation
Hyeonsu B Kang, David Chuan-En Lin, Nikolas Martelaro, Aniket Kittur, Yan-Ying Chen, Matthew K. Hong
Abstract
Nature often inspires solutions for complex engineering problems, but it is challenging for designers to discover relevant analogies and synthesize from them. Here, we present an end-to-end system, BioSpark, that generates biological-analogical mechanisms and provides an interactive interface for comprehension and ideation. From a small seed set of expert-curated mechanisms, BioSpark’s pipeline iteratively expands them by constructing and traversing organism taxonomies, aiming to overcome both data sparsity in expert curation and limited conceptual diversity in purely automated analogy generation. The interface helps designers recognize and understand relevant analogs to design problems using four interaction features. We conduct an exploratory study with design students to showcase how BioSpark facilitated analogical transfer of ideas but was limited in conveying active ingredients, the core abstraction underpinning how mechanisms work. We discuss this limitation and other implications such as generative hallucination that could facilitate shifts in human exploration of new design spaces.