Design representation as semantic networks

Design representation is a common task in the design process to facilitate learning, analysis, redesign, communication, and other design activities. Traditional representation techniques rely on human expertize and manual construction and are difficult to repeat and scale. Here, we present a methodology that utilizes a readily available large-scale multidisciplinary design knowledge base (KB) to automatically generate design representation as a semantic network, i.e., a network of the entities and relations, based on design descriptions in textual form. The methodology requires no ad hoc statistics, but a readily available KB. Thus, the KB has an essential impact on the usefulness and effectiveness of the methodology. Based on a participatory study, we observe the effectiveness and differences of the semantic network representations that are automatically generated with alternative KBs. Specifically, a KB that is trained on engineering-related data, TechNet, provides a more sensible representation of engineering design than commonsense KBs, WordNet and ConceptNet, to the participants who are engineers. We further discuss the implications of the findings and future research directions to enhance design representation as semantic networks. © 2022 Elsevier B.V.