Modelling the sensory characteristics of Scotch whisky using neural networks - a novel tool for generic protection

摘要

Scotch whisky is tightly defined by UK and EC legislation. It is vital to monitor products to ensure that they conform to this legislation. Sensory evaluation is an important aspect of generic authenticity. This involves appraisals by a highly trained whisky panel, in order to determine whether or not sensory attributes are consistent with those expected of a Scotch whisky. However, this approach is subject to bias due to personal preferences, and a more objective method of evaluation is required. This research describes the development of a neural network model, based on sensory profile data from 144 Scotch whiskies and Non-Scotch products, and its application as a tool for classifying products in terms of whether or not they have the sensory characteristics expected of a Scotch whisky. The improved performance of this approach is demonstrated for a range of Scotch whiskies, whiskies of other origins and non-whisky spirit drinks.