The model described here, which links olfactory semantic profiles to c
anonical similarities, performs significantly better than the current
state of the art, mainly due to the use of an optimization procedure.
Application of Tversky's model in this study shows that in olfaction,
estimation of similarities is based exclusively on the distinctive ele
ments of the profiles, and does not take into account the common eleme
nts. Moreover, the optimum number of descriptors to reckon with seems
to lie between 25 and 30.