A COMPLETE, FLEXIBLE FUZZY-BASED APPROACH TO THE CLASSIFICATION PROBLEM

We present an algebraic structure as a complete methodology of classif ication. Utilizing this structure, we apply an algebraic approximation to the problem of generating appropriate clusters of objects characte rized by fuzzy attributes More precisely, the values of the attributes are expressed in terms of linguistic labels, and thus are handled as fuzzy numbers. This opens new possibilities in all those fields for wh ich the need to describe the population under analysis by means of mor e natural terms becomes crucial. In fact, in these cases, the applicat ion of resolution strategies based on the adoption of ''standard'' met hods, such as a distance matrix, appears as a brutal effort to adapt q uantitative methods to qualitative problems.