Las listas de densidad semántica y la lexicografía funcional computacional

Abstract

Most researches conclude that the best method for building computational lexicons is the automatic analysis of machine-readable dictionaries. All the information needed for NLP lexicons can be found in a dictionary text. We have to decode the information presented by the human lexicographer in order to make it accessible in a computational way. In the transduction of a dictionary text into a Functional Grammar format we propose an enriched model of the basic component of S.C. Dik’s theory, the Fund. The meaning postulate associated to each predicative frame will take the form of a semantic density list, which enables the system to create a conceptually-motivated network connecting the predicates of the lexicon. This cognitive structuring of lexicon, also advocated by the Functional-Lexematic Model, is grounded on the Stepwise Lexical Decomposition approach. In this paper we therefore describe the automatic construction of semantic density lists for verbal predicates in Spanish.