Customers with diversified tastes have come to expect more than high product quality and performance. A high level of added value is required, and the provision for products that meet the individual requirements customers has become important. In recent years, delight design has attracted attention as a method for enhancing customer satisfaction. Delight design describes a design that offers attractive qualities beyond the required performance and quality requirements. However, attractiveness depends on the quality of the designer; furthermore, there are difficulties associated with the definition and expression based on similarities between the actual design and the original idea.
Therefore, in this study, we have proposed two methods for constructing neural network models: one for a value model using customer KANSEI data and one for an evaluation model for customers (designed as an inverse model). The customer value model creates the required product design from KANSEI. The evaluation model analyzes KANSEI for the individual customer and creates a design (delight design) according to KANSEI engineering and the preferences of the individual customer. The method proposed in this study was applied to the crack pattern seen on ceramic surfaces.