Abstract:
The paper is concerned with fuzzy relations of similarity between elements of a set for forecasting the membership functions determined on this set (functioning by analogy). In particular, the expert information on similarity (subjective similarity) obtained from analysis of object descriptions is made to agree in an optimal way with the critical (objective) similarity on the available sample by using a new formal description of information content in the descriptions of the elements. The proposed mathematical tools are applied to systems incorporating knowledge bases.