- From: <Peter.Hendler@kp.org>
- Date: Wed, 22 Aug 2012 07:04:37 -0700
- To: matthias.samwald@meduniwien.ac.at
- Cc: kerstin.l.forsberg@gmail.com, meadch@mail.nih.gov, public-semweb-lifesci@w3.org
- Message-ID: <OF6C5F54DA.0B3736B8-ON88257A62.004CEF41-88257A62.004D538C@kp.org>
Again, in real life as we do this, we are really talking about SNOMED with it's EL+ logic. It's like OWL without negation or disjunction. We only have SNOMED associated with a few nodes in a model, usually the ones that would indicate the diagnosis or what procedure was done. The idea of labels would be to make what we do explicit (we know what we do but how would you happen to know) and generalize it so that in the future if anyone needed another ontology for another purpose they could add it to their model. You mentions something I wonder about. If you lock the ontology (for example this is a release of SNOMED and you cant add new terms), and you run the reasoner and store all the inferred triplets as RDF along with all the stated ones, then could you really do just SPARQL at that point and not need a reasoner? NOTICE TO RECIPIENT: If you are not the intended recipient of this e-mail, you are prohibited from sharing, copying, or otherwise using or disclosing its contents. If you have received this e-mail in error, please notify the sender immediately by reply e-mail and permanently delete this e-mail and any attachments without reading, forwarding or saving them. Thank you. "Matthias Samwald" <matthias.samwald@meduniwien.ac.at> 08/22/2012 04:05 AM To <meadch@mail.nih.gov>, Peter Hendler/CA/KAIPERM@KAIPERM cc <kerstin.l.forsberg@gmail.com>, <public-semweb-lifesci@w3.org> Subject Re: seeks input on Study Data Exchange Standards Peter wrote: > Using both SQL and subsumption you can automatically find things like this: > "find all disorders that are a kind of adverse drug reaction where drug is a subtype of antibiotic and was given for a kind of gram negative bacterial infection of the digestive system". Simple subsumption such as that can be inferred by basic, RDFS-type reasoning. I don't see any potential problems caused by OWL's open world assumption here (please point them out if there are any). Indeed, the open-world assumption of OWL can make creating expressive ontologies and using reasoners tricky. However, I do not see why the same should be true for using RDF, basic RDFS subsumptions and SPARQL. Could you provide some examples? If we wanted to use more expressive ontologies with "intensional" entities (i.e., defined classes) in the overall system, we could simply run a reasoner and materialize the inferred statements for each ontology before it is 'shipped' for use by other systems. These other systems could then use simple RDF(S) and SPARQL (and maybe SPARQL rules), without the performance issues and potential unexpected consequences of open-world reasoning with expressive OWL ontologies. Specifying if and how exactly each specific node should be reasoned upon sounds so complex that I cannot imagine it working in any practical setting. Best, Matthias
Received on Wednesday, 22 August 2012 14:05:23 UTC