Referential modes of reasoning

The authors study models of referential structures and referential modes of reasoning for fuzzy data. The style of information processing considered is aimed at reasoning about some global properties of the spaces in which the fuzzy data are situated. The distributed models, designed in terms of logic-based neural networks, realize the mapping of these properties between the spaces. The scheme embraces reasoning about similarity, difference, dominance, and inclusion of the conclusions that is based on the corresponding relationships between the universes and their strength discerned for the antecedents. For instance, the conclusions issued within the scheme are of the form: b and b´ are similar, b and b´ are different, etc. It is shown that by considering the available degrees of satisfaction of these properties, the corresponding fuzzy sets of conclusion can be determined