Introducing Similarity Relations in a Framework for Modelling Real-World Fuzzy Knowledge

Abstract

There is no need for justifying the use of fuzzy logic (FL) to model the real-world knowledge. Bi-valued logic cannot conclude if a real-world sentence like "the restaurant is close to the city center" is true or false because it is neither true nor false. Letting apart paradoxes' sentences, there are sentences (as the previous one) that are not true nor false but true up to some degree of truth or true at least to some degree of truth. In order to represent the truth or falsity of such sentences we need FL. Similarity is a relation between real-world concepts. As in the representation of the truth of the first sentence, the representation of the similarity between two (fuzzy or not) concepts can be true, false or true up to (or at least to) some degree. We present syntactic constructions (and their semantics) for modelling such relation between concepts. The interest is in, for example, obtaining "spanish food restaurants" when asking for "mediterranean food restaurants" (only if the similarity between spanish and mediterranean food is explicitly stated in the program file). We hope this allows to represent in a better way the real-world knowledge, specially the concepts that are defined just by their similarity relations to some other concepts. © Springer International Publishing Switzerland 2014.