Rapid sensory profiles with DISTATIS and Barycentric Text Projection: An example with amari, bitter herbal liqueurs

Abstract

The sorting task is popular with sensory scientists because it provides rapid sensory profiles, even when product sets are complex or fatiguing. However, these sensory profiles generally cannot identify which sensory properties are driving subjects' perception of product similarities or differences—a critical task for the sensory analyst. This paper presents DISTATIS with Barycentric Text Projection as a solution that combines sorting-task and free-text data into a single-pass analysis to generate rapid, descriptive sensory profiles. This method is illustrated with a dataset generated by 25 subjects performing a replicated sorting and free-text description task in standard sensory-laboratory conditions on a set of 12 amari—bitter, herbal liqueurs that have not been previously analyzed in the sensory-science literature. DISTATIS with Barycentric Text Projection of the amari set produced sensory product maps that were readily interpretable. Using this analysis, the amari were grouped and described in ways that correspond to the available, popular literature descriptions of these products. The results of Barycentric Text Projection were compared to an independent Correspondence Analysis (CA) of the free-text data, and results were highly similar (RV=0.93). Future extensions of the method—such as analysis of descriptors in a check-all-that-apply (CATA) approach—are discussed. Overall, the successful, rapid, descriptive profiling of a set of 12 complex products using an untrained panel supports the potential of DISTATIS with Barycentric Text Projection as a sensory-evaluation tool.