Stream: ontology
Topic: Naming/repeating properties
Renato Iannella (May 19 2016 at 02:51):
When looking at the FHIR turtle, a native semantic web person would ask "why do the names of the properties include the Class and repeated so often?"
Properties are first-class objects in RDF/OWL so can (and do) exisit on their own. Their contextual use is declared with Class axioms.
For example there are about 29 fhir:*.patient properties
They are all about refering to the patient, so typically you would just have one patient property used in all the Class axions for BodySite, Claim, Encounter etc...
Grahame Grieve (May 19 2016 at 04:03):
ShEx - describing the shape of the resources. What other scope would it have?
Grahame Grieve (May 19 2016 at 04:04):
the names include the class - actually, the path, because the path is the unique name of the data element
Grahame Grieve (May 19 2016 at 04:05):
with regard to just having one property for references.... they are not all the same, even if they are all about referring to the patient
Grahame Grieve (May 19 2016 at 04:09):
although they share a bunch of commonality, they also have differences in context and relationships and definition - and sometimes, in constraints.
Renato Iannella (May 19 2016 at 23:52):
@Grahame Grieve RE: ShEx, just trying to find out what will be reperesented via ShEx (Section 1.17.3.1.6 of https://hl7-fhir.github.io/rdf does not give much away)...
Grahame Grieve (May 20 2016 at 00:04):
what is valid rdf for the resources - e.g. functionality equivalent to XML schema
Grahame Grieve (May 20 2016 at 00:04):
what else could it be?
Renato Iannella (May 20 2016 at 00:15):
So *all* Resouce validation will be via ShEx?
The current turtle includes: os:occurs os:Zero-or-many;
Renato Iannella (May 20 2016 at 00:27):
@Grahame Grieve RE: "share a bunch of commonality, they also have differences..."
I understand in an OO/UML world why you do that...in the SemWeb world things are a little different ;-)
https://www.w3.org/TR/sw-oosd-primer/
For example, in the classic Pizza Ontology http://protege.stanford.edu/ontologies/pizza/pizza.owl there is the single "hasTopping" property.
Its context/relationship/constraints are then expressed in the Pizza Class axions.
Eg The American Pizza hasTopping with {Mozza, Tomato, Peperoni} and the Capricosa Pizza hasTopping {Caper, Mozza, Tomato} etc...
What they don't do is create a AmericanHasTopping and CapricosaHasTopping properties (etc).
Grahame Grieve (May 20 2016 at 00:31):
validating the rdf representation of resources can be done by shex
Grahame Grieve (May 20 2016 at 00:31):
although there's lots of things to say that probably can't be said in Shex
Grahame Grieve (May 20 2016 at 00:32):
since any resource can be represented in rdf, any resource can be validated in ShEx. and it will probably validate a little more than schema or json-schema
Grahame Grieve (May 20 2016 at 00:34):
as for the bunch of commonality, if we had designed FHIR from the ground up for the semweb only, then we might end up there. but we didn't. and the list of things people want to do with RDF means that the differentiations we make need to made in the RDF too.
Grahame Grieve (May 20 2016 at 00:34):
I'd like to define the commonality in the RDF though
Grahame Grieve (May 20 2016 at 00:34):
so that AmericanHasTopping is also a hasTopping
Last updated: Apr 12 2022 at 19:14 UTC
