Similarity and prototype: Two key issues in perceptive and subjective information

Summary form only given. Subjective information is very difficult to elicit, to represent and to use because of its complexity. Among causes of subjectivity, perception is the most common and L.A. Zadeh has proposed a computational theory of perceptions, fuzzy set theory providing interesting tools to manage them. Imprecision of language and perceptions are closely related and U. Eco considers that concepts have the purpose of unifying various sensorial impressions. For Rissland, concepts are necessarily messy, with a core of typical elements and grey areas of interpretation. She considers that similarities are key notions to manage concepts. We present a general framework for similarities providing solutions to construct prototypes in a fuzzy setting, in agreement with classic works in psychology. We give some examples of their use for the management of perceptive and subjective information.