Needs-Based Product Configurator Design for Mass Customization Using Hierarchical Attention Network

Mass customization aims to provide goods and services that meet each individual customer's needs with a level of efficiency close to that of mass production. It is also a viable smart manufacturing strategy for companies that want to gain a competitive advantage in the current business environment. Product configurators are one of the major toolkits enabling mass customization. Existing product configurators require customers to choose from a set of predefined attributes or a list of component alternatives. However, customers may feel confused when configuring products if they do not have the necessary domain knowledge about the product. This article proposes a needs-based configurator mechanism that takes customer needs expressed in natural language as input to generate satisfactory product variants as output. This method leverages online product review data to distill the knowledge of customer preferences and needs, which then maps onto the product attribute specifications. A hierarchical attention network is applied to fully extract the information in the review text, which emphasizes the important keywords and phrases. We have obtained the promising experimental results, and our proposed needs-based configurators could help customers to find satisfactory product configurations with high recall rates. Note to Practitioners-Smart service is an indispensable element of smart manufacturing. It tries to offer customized solutions to address each customer's needs. Thus, customers need to be well integrated into the whole smart manufacturing cycle, specifically the initial design phase. Product configurators have been widely adopted in various industries to elicit customer needs and transform them into tangible product variants. Most existing configurators require customers to possess enough expertise to specify the components or attribute alternatives, which satisfy their needs. This poses challenges for customers to configure the product and for companies to realize the smart and mass customization strategy. This article proposes a needs-based configurator mechanism that takes customer needs expressed in natural language as input to generate satisfactory product variants as output. Deep-learning techniques are applied to implement the configurator. The new configuration mechanism could shield customers from the demanding process of specifying suitable attribute choices from a large pool and greatly improve the user experience.

Automated summary

The article proposes a needs-based configurator mechanism that takes customer needs expressed in natural language to generate satisfactory product variants. This approach helps customers find suitable product configurations with high recall rates and improves the overall user experience in smart manufacturing.