Abstract: An on-line recognition model of Kansei preference in product shape design, which can obtain unstructured information of customer in structured form，is put forward. Firstly, semantic differential and multidimensional scaling are used to complete mapping from customer′s Kansei space to space of Kansei designing. Secondly, an orthogonal design is used to reduce times of experiment. Thirdly, a shifting mode used to generate choiceset of orthogonal design. Finally, a discrete choice model is used to analyze data, which can obtain relationship between Kansei and designing elements. The example of car shape elements is taken to illustrate the model, of which exterior is comprehensible and reasonable, and interior is scientific and religious.

Key words:  product shape, kansei engineering, on-line, choice-based conjoint analysis