Identifying Styles in Product Design

Abstract

This paper introduces an approach to identify styles in product designs. We first built a theoretical foundation on what a style is and how designs are configured into styles. On this basis we executed a graph-clustering algorithm to identify styles in a population of design patents filed in the USA between the years 1976 to 2010. The design patent is uniquely suitable for this task because it is purely focused on how designs look. We validated the outcome obtained from the algorithm using experiments, which suggests that the algorithm outcome is no different from how humans would configure designs into styles. By establishing styles in designs, we open up a new avenue towards design research, upon which the dynamics of styles can now be studied.

Keywords: design, styles, Clustering Analysis, experiments