Kansei-mining: identifying visual impressions as patterns in images

Kansei engineering is a quite recent discipline aimed at understanding and modeling how subjective information is processed. We address a particular aspect of kansei engineering, which is visual impression, i.e. impressions experienced when observing images. They are highly subjective and difficult to explicit. We address the issue of how to reduce the complexity and uncertainty of the mapping between visual impressions and impression words used to convey them. We argue that the cooperation of the user is crucial in order to direct the system's modeling activity of this mapping. The role of the user takes the form of an externalization process aimed at transforming the user tacit knowledge into modeling hypotheses. We propose a conceptual space that supports the translation of these hypotheses into a formal specification to be then implemented in the computer. Cognitive maps are proposed as a tool to trigger the externalization process in the user and to support objective reasoning upon subjective beliefs.