par Magrez, Paul 
;Smets, Philippe
Référence International journal of intelligent systems, 4, 2, page (181-200)
Publication Publié, 1989-06
;Smets, Philippe Référence International journal of intelligent systems, 4, 2, page (181-200)
Publication Publié, 1989-06
Article révisé par les pairs
| Résumé : | Models for fuzzy deductive reasoning are becoming important in the field of knowledge-based systems. Zadeh suggested a compositional rule of inference, also called Generalized Modus Ponens. In order to study the fine structure of this model, we test seven rules of implication, usually found in the fuzzy literature, and two kinds of compositional rule with several properties which seems to be required by commonsense reasoning. But, as no rule is shown to satisfy these properties, we propose a new model for fuzzy Modus Ponens. The main idea of this model is to find a fuzzy extension of the meta-level rule of inference, in comparison with the object-level extension of the disjunctive syllogism such as the Zadeh's model. It is shown that our model satisfies all the intuitively required properties, and moreover is computationally very easy to apply. |
