Perfumery Radar: A Predictive Tool for Perfume Family Classification

Abstract

The classification of perfumes into olfactory families has been done for years on the basis of sensorial analysis or odor descriptors, but none of these methods has attained universal acceptance. In this work is presented a methodology called perfumery radar (PR) that predicts the classification of perfumes using the olfactive families that perfumers use. The PR introduces some scientific basis, reducing the arbitrariness of perfume classification to the empirical classification of pure odorants. The odor intensity of pure fragrances in a liquid mixture is predicted using the odor value concept, considering molecular interactions between components. Radar plots are used to represent olfactory families and transform quantitative information into qualitative. Perfumery radars have been obtained for several commercial perfumes and compared with existing experimental classifications. Another validation using headspace GC analysis was also performed with satisfactory results. It is shown that the PR methodology is able to predict the primary olfactive family of perfumes, according to the experimental classification given by perfumers. The prediction of secondary and tertiary families agreed with some of the empirical classifications in most cases, although there was little agreement among those at this level.