A nonadditive multiattribute evaluation model using Kansei data

This study deals with evaluation of products according to the Kansei, which is an individual subjective impression reflecting the aesthetic appeal of products. To do so, after introducing a probabilistic approach to generating Kansei profiles involving fuzzy uncertainty and underlying semantic overlapping, we have proposed a two-phase nonadditive multiattribute Kansei evaluation model based on probabilistic Kansei profiles. First, a target-oriented Kansei evaluation function is proposed to induce nonlinear Kansei satisfaction utility according to a consumer's personal Kansei preference, which provides a good description of the consumer's preference. Second, after formulating a general multiattribute target-oriented (MATO) Kansei evaluation function, a nonadditive MATO Kansei evaluation function is proposed based on an analogy between the general MATO Kansei evaluation function and the Choquet integral. The main advantages of our model are its abilities to deal with good description of personalized Kansei preferences as well as mutual dependence among multiple Kansei preferences.