Sensory characterisation using text mining analysis: An approach for a site-typicity description in Chilean Sauvignon blanc wines

Text mining can be a valuable tool for the sensory description of wines and word clouds provide a graphical representation of sensory descriptors generated by a sensory panel. Using these tools, the sensory characteristics of commercial Sauvignon blanc wines were analysed to assess the influence of three wine producing regions, Casablanca, Leyda, and San Antonio. The wines from San Antonio and Leyda showed higher levels of acidity, both chemically and sensorially, compared to those from Casablanca, which were characterised by greater complexity. Text mining analysis allowed a detailed examination of sensory descriptors to be carried out across the study regions and revealed Casablanca wines to feature a broader diversity of descriptors than those from Leyda and San Antonio. While some descriptors overlapped across the three regions, such as balanced and asparagus, distinctive sensory profiles were evident. The wines from the three regions were described as having greener and fresher notes, with descriptors such as vegetal, asparagus, and herbaceous, and a mouthfeel characterised by citrus notes and acidity. Using text mining analysis, it was possible to describe the typicity of Chilean Sauvignon blanc wines and link it to their specific terroir.