The present work proposes an alternative approach to represent approximate reasoning inferences, called approximate rules of inference, where a mechanism is introduced in order to define the syntactic expression for the approximate inferred conclusions. The evaluation of the approximate validity of the inference is done through: 1) the numerical weights attached to the logical formulae; 2) the explicit inclusion of the notion of semantic approximation/equivalence between fuzzy propositions in the process of inference; and 3) considering the set of generative axioms be complemented by a rule of detachment (Modus Ponens). It is possible to see from the approximate rules of inference's logical aspects and interpretations some potentialities of the present methodology to formalize and also to bridge the gap between the distinct approaches for existing approximate reasoning inferences.
展开▼