Fuzzy relational compositions, Excluding features, Fuzzy quantifiers, Classification, Odonata

This article focuses on fuzzy relational compositions, that unquestionably played a crucial role in fundamentals of fuzzy mathematics since the very beginning of their development. We followthe original works aiming at medical diagnosis, where the compositions were actually used for a sort of classification and/or pattern recognition based on expert knowledge stored in the used fuzzy relations. We provide readers with short repetition of theoretical foundations and two recent extensions of the compositions and then, we introduce how they may be combined together. No matter the huge potential of the original compositions and their extensions, if the features are constructed in a certain specific yet very natural way, some limitations for the applicability may be encountered anyhow. This will be demonstrated on a real classification example from biology. The proposedcombinations of extensions will be also experimentally evaluated and they will show the potential for further improvements.