In recent years, recommendation systems have been widely used for product recommendation at EC sites and so on. Existing recommendation systems are mainly based on cooperative filtering, but this approach simply estimates the customer preferences from the information about other customers with similar purchase histories and doesn’t take into account customer kansei, i.e., the degree of the impressions and preferences that they receive from the product, the design / aesthetic features of the product, and their corresponding relationships. For more accurate estimation of customer preferences, a new recommender system that takes into account customer kansei is proposed in this paper. The proposed system makes product recommendations by collecting information about the many different types of products that customers have purchased or preferred in the past and analyzing the correspondence relationships between the customer preferences and their design / aesthetics. In the case study, recommendation of a long wallet was made from the information about 6 types of customer’s favorite products (backpack, smartphone case, sneaker, pencil case, tie and scarf) for 18 subjects and it was confirmed that the proposal system can make recommendations with a certain degree of accuracy.
CITATION STYLE
Kobayashi, M., & Takeda, T. (2020). Product recommendation based on analysis of aesthetic elements used to customer’s favorite products. Computer-Aided Design and Applications, 18(4), 682–691. https://doi.org/10.14733/cadaps.2021.682-691
Mendeley helps you to discover research relevant for your work.