A key to fuzzy-logic inference (Q1209538): Difference between revisions

The paper is an analysis of the well-known problem of modus ponens and other classical rules of inference (modus tollens, reductio ad absurdum) in the setting of fuzzy logic. Without giving clear definition of a syntactical and semantical system of fuzzy logic, the author proposes to introduce rules of inference in fuzzy logic which are classified as `symbol elimination' and `symbol introduction'. The rules are, e.g., implication elimination (modus ponens) and introduction, negation, conjunction, and disjunction elimination and introduction, and also some extensions with quantifiers. Most space is devoted to modus ponens which is based on the implication of the form \[ (A\Rightarrow B)=(A'\vee B). \tag{1} \] In the discussion, the author refuses other kinds of implications usually considered in fuzzy logic including Łukasiewicz's one. A proposal to formalize also fuzzy mathematical induction and to solve the Yale's shooting problem using the presented system, is also given. The modus ponens is defined to give the resulting formula \(B=(A\Rightarrow B)\) if the truth of \(A\Rightarrow B\) is strictly greater than the negation \(A'\), and to give no result otherwise. Unfortunately, the paper does not reflect the most formally elaborated system of fuzzy logic based on residuated lattice of truth values, which was proved to be syntactico-semantically complete, and moreover, which is the only system with truth values filling the interval \(\langle 0,1\rangle\) to have this property (up to isomorphism). Within it, Łukasiewicz's implication works well and if one wants (1) as an alternative implication, it is possible to define the rule of modus ponens without the restrictions described above.