Skip to Main content Skip to Navigation
Journal articles

Generalized possibilistic logic: Foundations and applications to qualitative reasoning about uncertainty

Abstract : This paper introduces generalized possibilistic logic (GPL), a logic for epistemic reasoning based on possibility theory. Formulas in GPL correspond to propositional combinations of assertions such as “it is certain to degree λ that the propositional formula α is true”. As its name suggests, the logic generalizes possibilistic logic (PL), which at the syntactic level only allows conjunctions of the aforementioned type of assertions. At the semantic level, PL can only encode sets of epistemic states encompassed by a single least informed one, whereas GPL can encode any set of epistemic states. This feature makes GPL particularly suitable for reasoning about what an agent knows about the beliefs of another agent, e.g., allowing the former to draw conclusions about what the other agent does not know. We introduce an axiomatization for GPL and show its soundness and completeness w.r.t. possibilistic semantics. Subsequently, we highlight the usefulness of GPL as a powerful unifying framework for various knowledge representation formalisms. Among others, we show how comparative uncertainty and ignorance can be modelled in GPL. We also exhibit a close connection between GPL and various existing formalisms, including possibilistic logic with partially ordered formulas, a logic of conditional assertions in the style of Kraus, Lehmann and Magidor, answer set programming and a fragment of the logic of minimal belief and negation as failure. Finally, we analyse the computational complexity of reasoning in GPL, identifying decision problems at the first, second, third and fourth level of the polynomial hierarchy.
Complete list of metadatas

Cited literature [80 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-02640549
Contributor : Open Archive Toulouse Archive Ouverte (oatao) <>
Submitted on : Thursday, May 28, 2020 - 2:59:08 PM
Last modification on : Tuesday, January 26, 2021 - 4:14:02 PM

File

Dubois_22163.pdf
Files produced by the author(s)

Identifiers

Citation

Didier Dubois, Henri Prade, Steven Schockaert. Generalized possibilistic logic: Foundations and applications to qualitative reasoning about uncertainty. Artificial Intelligence, Elsevier, 2017, 252, pp.139-174. ⟨10.1016/j.artint.2017.08.001⟩. ⟨hal-02640549⟩

Share

Metrics

Record views

18

Files downloads

54