Design-by-Analogy: Effects of Exploration-Based Approach on Analogical Retrievals and Design Outcomes

This paper presents VISION (Visual Interaction tool for Seeking Inspiration based on Nonnegative Matrix Factorization), a computational design-by-analogy (DbA) tool that enables designers to visually explore a space of analogical inspiration for creative idea generation. While many currently available DbA tools use a query-based approach for retrieving analogies (i.e., input keywords or functions to return a set of relevant results), VISION allows designers to retrieve a collection of design analogies that are related to topics of interest and explore a space of potential inspiration, the way one would gather books of particular topics from multiple shelves at the library to find potential resources. Two cognitive engineering design studies were conducted to evaluate the efficacy of VISION during the conceptual design process. In the first study, conducted in a controlled-lab setting, VISION was evaluated based on its effect on the quantity, quality, novelty, and direct physical similarity ratings of design outcomes. In the second study, conducted in a graduate engineering design class, VISION was evaluated based on designers’ abilities to retrieve analogies from different domains and analogies that are different from already existing design solutions. Studies show that VISION could provide an alternative to the query-based search that many DbA computational support systems use and open up new opportunities for designers to benefit from computationally supported analogies.