FUZZY SET THEORY APPLIED TO PRODUCT CLASSIFICATION BY A SENSORY PANEL

. It is frequently impossible to meet the assumptions underlying the statistical approach to classification of food products by a sensory panel. To find an alternative, we have investigated the applicability of the fuzzy set theory. Within a fuzzy set framework it is acceptable that a product belongs to several classes simultaneously and no assumptions regarding the distribution of sensory properties for a product class are made. Fuzzy classification models can be constructed from a set of training objects by linking the soft class labels to the sensory attributes applying an inference procedure based on fuzzy logic. A number of fuzzy inference procedures has been evaluated using a number of attribute sets. A satisfactory classification has been found using a very simple implication rule and a set of three attributes.