From analogy to innovation: a creative conceptual design approach leveraging large language models
Author(s)
Type
Journal Article
Abstract
Integrating creative concepts into Product Design and Manufacturing Systems (PDMS) is important for product innovation. However, current PDMS lack cognitive capabilities, particularly in reasoning and synthesis, which are essential for conceptual design. As a result, designers face challenges in retrieving relevant analogies, establishing meaningful mappings, and integrating knowledge into new design concepts. This paper proposes a computational conceptual product design approach that integrates Large Language Models’ (LLMs) knowledge representation with an analogy-based structured retrieval mechanism, supporting designers to explore and recombine design patterns and functionalities in an intuitive manner. Benefiting from the zero-shot learning and prompting capabilities of LLMs, given a source domain, this approach allows reasoning target domain based on abstract correspondences in both morphological and semantic associations. A template for the combinational regulation of the reuse of analogical knowledge has also been formulated. By decomposing analogical knowledge into ontological distinction, inspirational feature recognition, and associative mapping explanation, a creative conceptual design stimulation path is formed. An interactive tool named ViMimic based on this approach has been developed through a case study with 18 participants. Evaluation results demonstrate that the approach improves creative performance, increasing the novelty and functionality of conceptual designs by 51% and 22% respectively according to expert evaluations. It also boosts the efficiency and diversity of analogy mapping by 30% based on objective measures, while enhancing creative experiences and reducing cognitive load as measured in the participants’ self-assessment.
Date Issued
2025-09-01
Date Acceptance
2025-04-28
Citation
Advanced Engineering Informatics, 2025, 67
ISSN
1474-0346
Publisher
Elsevier BV
Start Page
103427
End Page
103427
Journal / Book Title
Advanced Engineering Informatics
Volume
67
Copyright Statement
Copyright © 2025 Elsevier Ltd. This is the author’s accepted manuscript made available under a CC-BY licence in accordance with Imperial’s Research Publications Open Access policy (www.imperial.ac.uk/oa-policy)
License URL
Publication Status
Published
Article Number
103427
Date Publish Online
2025-06-05