A descriptive algorithm for a wine tasting lexicon corpus

Abstract

The aim of this paper is to show advances in testing the validity of a procedure of computational identification of the components that make up the meaning of expressions in the restricted sub-domain of wine tasting notes (WTN). This takes the form of a proposed linking algorithm, which includes a number of tagging components. These components range from non linguistic ones, with taggers for perceptual input and world knowledge to traditionally linguistic ones, such as parsers or dictionary description. The proposed methodology is the Clashing Identification Procedure (CIP). The CIP is presented as a procedure, where different types of lexical descriptors produce a clashing when combined in a string of components that take the form of an algorithm. This leads to a subsequent reduction of the corpus until a manageable size of it is reached. The interest of designing tagging components for a WTN corpus lies on its contribution to the identification of metaphoric and synaesthetic expressions that are frequently used in them and to other disambiguating tasks. That is, how to computationally deduce information relevant to the construction of metaphors, and how the contribution of such design will help connecting linguistic and encyclopaedic knowledge.