No label defined (Q4944944)

When we move from a crisp (2-valued) theory to its fuzzy version, we often face several choices, because formulations which are equivalent in 2-values logic often lead to different fuzzifications. A natural question is: Which fuzzification is the most natural? The author provides a general answer for the situations when we want to generate a fuzzy version of a \textit{logic} (e.g., of modal logic). The general idea behind this answer is a notion of ``weaving'' of two logics \(L_1\) and \(L_2\): we start with statements from \(L_1\), then we combine these statements by allowing formulas from \(L_1\) as atoms in \(L_2\); then, we use the resulting formulas as atoms of \(L_1\), etc. A natural fuzzy extension of a logic \(L\) can thus be defined as the results of weaving \(L\) and the standard fuzzy logic \(L_0\). Interestingly, we can apply this fuzzification idea to fuzzy logic \(L_0\) itself, and get a variant of ``second-order'' fuzzy logic, in which statements of the type ``the degree of truth in \(A\) is equal to \(r\)'' are also true to some extent, and thus, the degree of truth in \(A\) is described not by a single number \([0,1]\), but by a fuzzy subset of \([0,1]\). For the entire collection see [Zbl 0927.00008].