Flavours Correlation Determination by Non Supervised Neural Network in Virgin Olive Oil

Olive Oil is the first food product for which the assessment of the organoleptic quality by means of the Panel Test (PT) was introduced at the level of European Union legislation as a criterion for the sensory evaluation in the Reg. EU 1348/2013 annex XII (revision of Reg. EEC 2568/91). The standard establishes that olive oil must be tasted according to the rigid standards and regulations of the Panel Test (PT) that is carried out by a group of at least 8 professional tasters with a chief, that, through the tasting, give a numerical subjective evaluation (from 0 to 5) to many questions about flavours, in a questionnaire. The ensemble of the scores, reached by each components, defines the organoleptic quality. The amount of these numbers suggests, for better readability, the use of an unsupervised Artificial Neural Network (ANN), which identifies the best two-dimensional representation of the samples that have stimulated the PT. Analysing the coefficients relating a flavour to output neurons of the Kohonen Map, it is possible to individuate the heaviest. If it happens that more heavier coefficients inherent to different flavours, flights the same neuron, an internal correlation will be searched.