No label defined (Q4944946)

A general (crisp) logic is sometimes defined as a pair \((F,{\mathcal D})\), where \(F\) is a set of formulas, and \(\mathcal D\) is a closure operator with the following meaning: \({\mathcal D}(S)\) is the set of all formulas that can be deduced from \(S\). This definition is based on the idea that knowledge can be represented by describing which formulas from \(F\) are known to be true, i.e., by a (crisp) set \(S\) of formulas which are known to be true. (It is sufficient to describe what we know to be true since knowing that a formula \(f\) is false is equivalent to knowing that \(\neg f\) is true.) Fuzzy logic takes into consideration that in real life, we may have only \textit{partial} confidence in some statements \(f\in F\). Thus, knowledge is better described as a \textit{fuzzy} subset of the set \(F\). Many properties of classical deductive systems can be naturally generalized to such fuzzy sets, because for these properties the only operations with sets that we really use are the union and intersection, and in many proofs the only properties of these two operations which are used in these proofs are that these two operations form a lattice -- and fuzzy sets also form a lattice under union and intersection. It is therefore natural to define a general deductive system as a pair \((L,{\mathcal D})\), where \(L\) is a lattice, and \(\mathcal D\) is a closure operator on \(L\). The author shows that a large part of traditional fuzzy logic formalism can be reformulated in these new terms. This reformulation reveals an (often unexpected and stimulating) relationship between fuzzy concepts and concepts of traditional logic. The same general formalism can also deal with fuzzy constraints. For the entire collection see [Zbl 0927.00008].