Conference article

This study uses an Interactive Genetic Algorithm (IGA) for eliciting users’ perceptions of products. It aims to understand perceptions; and to determine the product’s attributes that contribute to reinforce - or inhibit - a given semantic dimension. The product proposed to illustrate the study; defined in collaboration with the Renault Company; is a digital instrument panel; integrated in a car dashboard; and the semantic dimension considered is its “sportiness”. After a parameterization of the instrument cluster; an interactive assessment test; based on IGA; is conducted on different products’ pictures with a panel of 30 participants. The IGA test proposes a set of 8 parameterized designs; which are iteratively presented to each participant with their pictures via an interactive interface. From these designs; the user has to select the most representative ones according to the considered semantic dimension (sportiness). From iterative choices of the user; the results show that the design of the product converges toward representative designs of the semantic dimension. We present in the paper the analysis of the results of the IGA test; using Hierarchical Ascendant Classification (HAC); univariate and multivariate analysis. The effect of the design variables and their different modalities on the sportiness of the product is discussed. The agreement between the different participants is also studied; to uncover different typologies of products and identify design trends. The results show that IGA can be an interesting alternative to classical rating tests on experimental designs; used in conjoint analysis or in Kansei Engineering.

Interactive genetic algorithms; product semantics; user centered design; multivariate analysis