Formalisation of at-line human evaluations to monitor product changes during processing - Integration of human decision in the dry sausage ripening process

In the food industry, information captions are affected by the scarcity of suitable on-line sensors especially with respect to sensory properties. Therefore it is often necessary to rely on human evaluations made close to the line, in order to assess quality characteristics of food products during their manufacture. This paper presents a method to collect and formalise the know-how of skilled operators concerning these evaluations. The classical descriptive sensory method has to be adapted to at-line measurement constraints. These specificities led us to develop the concept of sensory indicator and establish a new formalisation grid. This grid is composed of seven elements. Five come from the classical descriptive sensory method: name, definition, operating conditions, scale, and references as scale anchors. Two new elements focusing on product variability during processing have been introduced: a spatial characteristic that takes into account the spatial variability of products, and a temporal characteristic that takes into account changes of the product through time. The method is illustrated by an industrial application, i.e. the ripening stage during the manufacture of the dry sausage, Seven indicators have been identified and formalised using the grid. The results have shown that the evaluation of the operators is reliable, with reproducibility and discriminative ability being assessed. We consider that it is possible to formalise and transmit knowledge concerning at-line human evaluations of products and use them as input variables in process control models.