Análise de fatores paralelos (PARAFAC) na extração de informações de análise sensorial

Abstract

The food industry is constantly creating new products and making modifications to existing products, boosting market sales, and generating high revenues. These changes usually occur to improve the nutritional profile of the product and the cost of its production. For this to occur smoothly, it is the responsibility of food companies to ensure safety and quality, but also consumer acceptance and satisfaction with the product, since the consumer is the one who decides whether the product will be successful or not. Sensory analysis is the methodology used to assess sensory attributes and acceptability of these products, through interactions perceived by the senses of sight, taste, touch, smell, and hearing of each evaluator. Interpretations of sensory analysis results commonly occur by conventional statistics in a one-dimensional way. However, other multidimensional methodologies can also be used. The use of chemometric tools allows for greater interactivity between data, making it possible to correlate product, attribute, and taster results with multi-way data treatment tools. Thus, the objective of this work was to seek to extract information from sensory results exploring this greater interactivity between products, sensory attributes evaluated, and their respective tasters. Therefore, results obtained by tasters trained in descriptive sensory analysis of cachaça and coffee drink, and results obtained by untrained tasters for affective hedonic sensory analysis of household use (HUT) in vegetable oils, were evaluated using the multivariate factor analysis tool parallels (PARAFAC). The results show that the tool allowed us to conclude that aromatization can bring characteristics to the cachaça so that it presents similar notes for some sensory attributes when compared to cachaça aged in wooden barrels. It was also shown that the tool allows to distinguish coffee with higher quality from sensory results. Furthermore, the agreement of tasters in the evaluation of samples of the same type of coffee can be seen as a measure of precision in the level of repeatability. For vegetable oils, the application of PARAFAC made it possible to verify that color was the attribute that most contributed to the affectivity and distinction of different oils, while sex and age of the tasters did not influence the preference for a certain oil.