Causal models have no complete axiomatic characterization

Simple item page
dc.creatorLi, Sanjiang
dc.date2008-04-15
dc.date.accessioned2026-07-07T09:32:42Z
dc.date.available2026-07-07T09:32:42Z
dc.descriptionMarkov networks and Bayesian networks are effective graphic representations of the dependencies embedded in probabilistic models. It is well known that independencies captured by Markov networks (called graph-isomorphs) have a finite axiomatic characterization. This paper, however, shows that independencies captured by Bayesian networks (called causal models) have no axiomatization by using even countably many Horn or disjunctive clauses. This is because a sub-independency model of a causal model may be not causal, while graph-isomorphs are closed under sub-models.
dc.description7 pages, 1 figure
dc.identifierhttps://arxiv.org/abs/0804.2401
dc.identifierhttp://arxiv.org/abs/0804.2401
dc.identifier.urihttp://salesiana.dossiersoluciones.com/handle/123456789/158876
dc.subjectArtificial Intelligence
dc.subjectLogic in Computer Science
dc.titleCausal models have no complete axiomatic characterization
dc.typetext

Files

Collections

arXiv