Automating customer needs to engineering characteristics mapping in quality function deployment: A deep learning approach

Quality Function Deployment (QFD) stands as a widely utilized toolkit in product development. It offers a systematic approach to analyze customer requirements (CRs) and convert them into engineering language for product design and production. By separating CRs and engineering characteristics (ECs), QFD ensures that the final product considers both technical aspects and customer usability. Despite its historical success, QFD encounters challenges in today's business landscape. The process is typically intricate, demanding in labor, and time-consuming. QFD teams, comprising customers, designers, engineers, marketing experts, and moderators, heavily rely on the experience and expertise of designers and engineers. Decision-making is integral to the QFD process, making it less adaptable to the current fiercely competitive and time-sensitive business environment. This paper introduces a streamlined QFD method, designed to automate the process rapidly and require fewer resources from companies. Leveraging extensive online product review text, our smart QFD method deduces the relationship between CRs and ECs. The application of Graph Convolutional Network (GCN) aids in extracting features for the CRs-ECs mapping, facilitating the QFD process. Experimental results demonstrate the effectiveness of the GCN-based QFD structure. Copyright © 2024 IEEE.