The tasting of beers (like other foods) is based on a description of flavors and recognition of sensory patterns that requires inherently personal and difficult to manage sensory skills. Reliability is therefore affected by the subjectivity of decisions, the training of people and a very rigorous harmonization of criteria. The idea of having specific sensors (state-of-the-art electronic nose / tongue) combined with advanced data processing systems that could become capable of objectifying perceptions would clear up doubts and uncertainties.
The field of application in the first instance would be to work with the complex sensors and data treatment that would give us notice of possible off-types once this system had learned to compare whether a sample fits with a beer with alcohol standard considered as reference obtained from having “trained” the system with a large number of samples that included the maximum reasonable variation. We also propose to work with a concept of “sensory distance” in relation to this pattern in order to establish decision or warning guidelines.
It is possible to separate the proposed solution into two parts, as long as they are uniquely validated:
– Specialized sensors for the recognition of physical-chemical parameters of beer
– AI: capable of training and recognizing with a sufficient approximation, the sensory differences between different types and varieties of beer.
Solutions not of interest
Conventional methods that require manual parameterization or are unable to learn and manage a large number of data that become useful information for the distinction of references of different beers.