Fuzzy Set Similarity Between Fuzzy Words

Abstract

Fuzzy set similarity measures determine the similarity between two fuzzy sets. Semantic similarity measures determine the similarity between concepts within an ontology. Research in determining sentence similarity has used semantic similarity between fuzzy words that have been structured in an ontology based on data provided by human experts to create fuzzy sets for these fuzzy words. The research uses four different kinds of fuzzy set similarity measures, three that are standard ones for type-1 fuzzy sets and another one based on the distance between defuzzified and normalized COGs for type-2 fuzzy sets and examines both the Pearson and Spearman correlations among these different fuzzy set similarity measures on fuzzy words from four of the six categories established in past sentence similarity research. The results show that all standard fuzzy set measures are highly correlated though some categories of fuzzy words result in higher correlations than others. The standard fuzzy set similarity measures have lower correlation with the one developed for similarity between type-2 fuzzy sets using their defuzzified and normalized COGs.