The role of synonymy and antonymy in 'natural' Fuzzy Prolog

Abstract

The aim of this paper is to attempt a first approach to a kind of 'natural Fuzzy Prolog' based on the linguistic relations of synonymy and antonymy. Traditionally, Prolog was associated to the clausal logic, a disposition of the classical logic in which the goals are conjectural theorems and the answers, provided by the interpreter, are achieved using resolution and unification. Both resolution and unification are the core of a Prolog interpreter. Classical Prolog has had and still currently has interesting applications in domains as natural language processing where the problems are verbalized using crisp language and algorithmic style. But as Zadeh pointed out, natural language is essentially ill-defined or vague. Fuzzy Prolog provides tools for dealing with tasks that involve vague or imprecise statements and approximate reasoning. Traditionally, fuzzy Prolog was related with the specification of facts or rules as a matter of degree. Degrees adopted several forms: single degrees, intervals of degrees and linguistic truth-values, represented by triangular or trapezoidal numbers. Fuzzy solutions using degrees are valuable, but far from the way employed by human beings to solve daily problems. Using a naive style, this paper introduces a 'natural fuzzy Prolog' that deals with a kind of natural resolution applying antonymy as a linguistic negation and synonymy as a way to match predicates with similar meanings. © 2012 Springer-Verlag Berlin Heidelberg.