Many-valued first-order logics with probabilistic semantics
This paper presents a probabilistic semantics of \(n\)-valued first-order logics that lies between the purely probabilistic semantics and the truth-functional semantics of \(n\)-valued Łukasiewicz logics. Purely probabilistic \(n\)-valued semantics formalize the idea of combining information obtained by \(n-1\) experts in a two-valued classical first-order logic. A closed formula \(\alpha\) gets the value \(k/(n-1)\) if \(k\) of the \(n-1\) experts consider the formula \(\alpha\) to be true. Note that in general, the truth value of the connectives (conjunction, disjunction, implication), cannot be expressed as a function of the truth values of its components. To formalize this idea, one considers an interpretation \({\mathcal I}=(D,\pi_1,\ldots,\pi_{n-1})\) that consists of \(n-1\) interpretations \((D,\pi_1),\ldots,(D,\pi_{n-1})\) over the same domain \(D\) such that \(\pi_1(f)=\ldots=\pi_{n-1}(f)\) for all function symbols of the language. In the purely probabilistic semantics (\(\text{Pr}_n\)), for each variable assignment \(\sigma\) the truth value \({\mathcal I}_\sigma(F)=k/(n-1)\), where \(k\) is the number of ``true values over \(n-1\) interpretations. Note that this interpretation preserves the classical truth values \(0\) and \(1\) respectively for closed contradictory and tautological formulas. In the Łukasiewicz semantics \(({\L}_n)\) only the atomic formulas are evaluated in that way and compound formulas are evaluated inductively according to the Łukasiewicz definitions (\({\mathcal I}_\sigma(\neg F)=1 - {\mathcal I}_\sigma(F); {\mathcal I}_\sigma(F \wedge G) = \min({\mathcal I}_\sigma(F),{\mathcal I}_\sigma(G)), {\mathcal I}_\sigma(\forall x F)= \min\{{\mathcal I}_{\sigma[x/d]}(F)|d\in D\}\)). For this interpretation, contradictions and tautologies are not preserved. A new semantics (\({\L}_n^{*}\)) that preserves this property is obtained by requiring for an interpretation \(\pi_1(p) \supseteq \pi_2(p)\supseteq \ldots \supseteq \pi_{n-1}(p)\) for all predicate symbols \(p\), and defining \({\mathcal I}_\sigma(F)\) as in the purely probabilistic semantics. It is shown that this interpretation preserves the Łukasiewicz definitions for formulas \(F\) and \(G\) without logical connectives. Moreover it is shown that this semantics has interesting computational properties. More precisely, \(n\)-valued logical consequence in disjunctive logic programs with \(n\)-valued disjunctive facts, can be reduced to classical logical consequence in \(n-1\) layers of classical disjunctive programs. Finally, it is shown that some deduction problems in \(n\)-valued logic programs have the same computational complexity as their classical counterparts. For the entire collection see [Zbl 0913.00028].
- Boolean-like and frequentistic nonstandard semantics for first-order predicate calculus without functions
- Generalization of the fundamental theorem of probability logic in multi-valued propositional logic
- Models for Many-Valued Probabilistic Reasoning
- scientific article; zbMATH DE number 2042624 (Why is no real title available?)
- Many-valued disjunctive logic programs with probabilistic semantics
- Factor semantics for n-valued logics
This page was built for publication: Many-valued first-order logics with probabilistic semantics
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4263812)
