Conference Dates

September 8-13, 2019


Biomimetics, the application of mechanisms observed in nature to inform technical solutions, is inherently cross-disciplinary. For the most part, however, practitioners are only expert in one domain, e.g., engineering. Being a layman in the other domain, biology, can make it hard and time-consuming to find and understand relevant information. Computer-Aided Biomimetics (CAB) involves the development of computational tools to overcome this domain-expertise mismatch.

Finding a bridge between engineering and biology has been challenging. Although a plethora of methodological approaches have been proposed to bridge the engineering and biology domains, Biomimetics remains adventitious and research intensive. We give an overview of previous research efforts in CAB and motivate our approach that revolves around the resolution of biological trade-offs. This is a unique approach, as previous work has always aimed to extract engineering functions from biological texts. We describe our novel CAB system that extracts trade-offs, a within-domain concept to indicate a dialectical relation between two or more biological traits.

We provide a description of our dataset for the extraction of trade-offs from biology research papers, as well as our state-of-the-art Relation Extraction system. The dataset consists of over 1.5k sentences taken from biology research papers, describing a trade-off or similar high-level relation between two or more concepts. Furthermore, we provide an in-depth analysis of the information extracted by our CAB system from a corpus of 10k biology research papers. We show in a qualitative analysis that our system extracts key concepts and relations from biology research papers that are relevant to Biomimetics.

Unique to our approach is that our system makes it feasible to collect a comprehensive list of the system parameters and solution principles used in biology. This enables statistical analysis, such as finding the distribution of fundamental principles among the resolution of various trade-offs. Notably, the solutions to trade-offs differ little over various hierarchical levels of biology. This makes our finding relevant to any research that aims to find desired, but underutilized, properties observed in nature.

Included in

Engineering Commons