No label defined (Q4250045)

In classical two-valued logic, the following ``rule of syllogism'' holds: if \(A\) implies \(B\) and \(B\) implies \(C\), then \(A\) implies \(C\). This paper analyzes when a similar rule holds in fuzzy logic. This question is not easy even to formulate, because in fuzzy logic, the statement ``\(A\) implies \(B\)'' is not necessarily absolutely true or absolutely false; in addition to these two classical truth values 1 (absolutely true) and 0 (absolutely false), the implication statement can have an intermediate truth value from the open interval \((0,1)\). In these terms, the above question can be reformulated, crudely speaking, as follows: if we know the truth values \(t_{AB}\) and \(t_{BC}\) for the implications ``\(A\) implies \(B\)'' and ``\(B\) implies \(C\)'', can we deduce a lower bound \(b_{AC}\) on the truth value of the implication ``\(A\) implies \(C\)'' in such a way that when \(t_{AB}\) and \(t_{BC}\) tend to 1, we will have \(b_{AC}\to 1\)? The author gives a necessary and sufficient condition for the existence of such a lower bound, and shows that these conditions hold for many implication operations used in applications of fuzzy logic (such as fuzzy control). For the entire collection see [Zbl 0910.00016].