Re: Detecting cycles in OWL
Franconi Enrico <[email protected]> Tue, 5 May 2020 07:44:51 +0000
| Newsgroups | gmane.org.w3c.semantic-web |
|---|---|
| Message-ID | <[email protected]> |
Il giorno 5 mag 2020, alle ore 02:54, Peter F. Patel-Schneider <[email protected]<mailto:[email protected]>> ha scritto: Yes, yes, for any particular OWL derivation you could write a SPARQL CONSTRUCT query that mimics that proof, but there are an infinite number of these derivations. More specifically, for each of the infinitely many OWL ontology structures, there are infinitely many SPARQL CONSTRUCT queries, which compute the (infinitely big) deductive closure for that OWL ontology structure. Some of these SPARQL CONSTRUCT queries may be infinite themselves. And even if we are happy to have only an incomplete version of the reasoner, we need a complete reasoner to find and/or validate the finitely many SPARQL CONSTRUCT queries. Also, I'm not sure that SPARQL CONSTRUCT queries are able to represent arbitrary graph transformations needed to express all OWL derivations; it may be the case but it should be proved. cheers --e.