This chapter presents several methods based on self-organized dissimilarity or similarity [54]. The first uses fuzzy clustering methods while the second is a hybrid method of fuzzy clustering and multidimensional scaling (MDS) [18], [31]. Specifically, a self-organized dissimilarity (or similarity) is defined that uses the result of fuzzy clustering where the dissimilarity (or similarity) of objects is influenced by the dissimilarity (or similarity) of the classification situations corresponding to the objects. In other words, the dissimilarity (or similarity) is defined under an assumption that similar objects have similar classification structures. Through an empirical evaluation the proportion and the fitness of the results of the method, which uses MDS combined with fuzzy clustering, is shown to be effective when using real data. By exploiting the self-organized dissimilarity (or similarity), the defuzzification of fuzzy clustering can cope with the inherent classification structures.