See also: IRC log
https://app.box.com/s/k7i68oyi6vg54pk8ebq8leue4e1v3j83
dbooth: discuss Michael's slides this afternoon
http://wiki.hl7.org/index.php?title=201605_Declarative_Mapping
Ken Lord's topic he wishes to discuss:
https://lists.w3.org/Archives/Public/public-semweb-lifesci/2016Apr/0004.html
thomas: concerns about a FHIR-specific mapping language
dbooth: Objections to giving Ken a brief time slot in Monteal?
harold: i put some ShEx examples
on the site
... Some of grahame's examples do not match the spec.
... Also need to get eric to put these into the ShEx
validators
<hsolbrig> https://github.com/w3c/hcls-fhir-rdf/tree/gh-pages/examples
dbooth: Next steps for ShEx examples?
harold: Need to work with grahame on references.
<hsolbrig> http://hl7-fhir.github.io/medicationexample1.ttl.html
<hsolbrig> http://hl7-fhir.github.io/allergyintolerance-example.ttl.html
<hsolbrig> https://github.com/w3c/hcls-fhir-rdf/blob/gh-pages/examples/AllergyIntolerance.shex
The JSON looks like this:
[[
"patient": {
"reference": "Patient/example"
},
]]
tony: In the shex, there should be one more level of reference. A fhir:Reference has a reference property and a display property: http://hl7-fhir.github.io/references.html
<hsolbrig> https://github.com/w3c/hcls-fhir-rdf/blob/gh-pages/examples/Reference.shex
<hsolbrig> Differences in Grahame's model -- sb. fhir:Reference.link <uri> vs. fhir:refernce
tony: we added a fhir:Reference.link <http://....> instead of grahame's fhir:reference <http://...> to point to the referenced object.
<ericP> <Condition xmlns="http://hl7.org/fhir"> <patient> <reference value="Patient/example"/> </patient> <asserter> <reference value="http://fhir.example/Patient/example"/> </asserter>
<ericP> </Condition>
<hsolbrig> Michael VDZ argues for leaving it on the top namespace
<hsolbrig> EricP: WHy do we have something different in RDF than in JSON XML
michael: Better to put it at the top level of the fhir: namespace, like grahame does it: fhir:reference .
<ericP> <Condition xmlns="http://hl7.org/fhir">
<ericP> <patient> <reference value="Patient/example"/> </patient>
eric: What if we use a relative URI?
<ericP> <asserter> <reference value="http://fhir.example/Patient/example"/> </asserter>
<ericP> </Condition>
eric: Is a FHIR processor
expected to recognize http://fhir.example/Patient/example
and Patient/example as referring to the same resource?
... Will it treat them as identical? Will it try to serialize
them the same way?
... If there is an expectation that everything is resolved
against the base, then those would be serialized the same
way.
dbooth: Need to be able to
round-trip
... Anything in double-quotes is a literal in RDF.
<hsolbrig> I have to leave - I've got another meeting. Will see you at the afternoon meeting today...
dbooth: because of the
round-tripping requirement, I think we need to maintain the
relative URIs
... Canonical FHIR JSON does not define a way to normalize
references; therefore relative URIs need to stay relative and
abs URIs need to stay abs.
ISSUE: Do we need a fhir:reference (name TBD) whose value is an RDF URI of the referent?
<trackbot> Created ISSUE-23 - Do we need a fhir:reference (name tbd) whose value is an rdf uri of the referent?. Please complete additional details at <http://www.w3.org/2014/HCLS/track/issues/23/edit>.
ISSUE: Must absolute and relative references be round tripped as is?
<trackbot> Created ISSUE-24 - Must absolute and relative references be round tripped as is?. Please complete additional details at <http://www.w3.org/2014/HCLS/track/issues/24/edit>.
eric: i.e., do we need to
maintain the distinction between relative and absolute
references?
... when round-tripping?
<scribe> ACTION: dbooth to ask Grahame and Lloyd [recorded in http://www.w3.org/2016/04/26-hcls-minutes.html#action01]
<trackbot> Created ACTION-54 - Ask grahame and lloyd [on David Booth - due 2016-05-03].
Who can make it? dbooth, harold, eric, rob
https://app.box.com/s/k7i68oyi6vg54pk8ebq8leue4e1v3j83
michael: I don't know what the
@base URI should be
... Might not have one if FHIR is being submitted
... Also fhir-net-api puts 'Component' in the names
dbooth: RDF does not have relative URIs, but certain RDF serializations do. in contrast, the FHIR model and FHIR JSON and FHIR XML do have relative URIs.
michael: rdfDotNet always generates xsd: prefix instead of xs: .
dbooth: We decided to standardize on using xs:
michael: Question about bundle reference. Should it just point to the referenced MedicationOrder object, instead of embedding it?
dbooth: how will it impact round tripping?
<mvdzel> bye have to go, will add some more examples to the slides and tty in Montreal
ADJOURNED
<hsolbrig> <PatientReference> { a [fhir:PatientReference]?, fhir:uri.id @<id>?, fhir:Reference.link @<Patient>?, fhir:Reference.reference @<uri>? }
<hsolbrig> <PatientReference> { a [fhir:PatientReference]?, fhir:Element.id @<id>?, fhir:Element.extension @<Extension>*, fhir:Reference.link @<Patient>?, fhir:Reference.reference @<string>? fhir:Reference.display @<string>? }
<ericP> http://piratepad.net/ericP
# Grahame is generating:
fhir: AllergyIntolerance.patient
[
... l <http://hl7.org/fhir/Patient/example>;
... Reference.reference [ fhir:value "Patient/example" ]
];
AGREED: Change fhir:reference to fhir:link to avoid confusion.
AGREED: We prefer fhir:link (at top level of FHIR namespace) instead of fhir:Reference.link, because fhir:Reference.link misrepresents itself as a property of fhir:Reference
# Grahame should be generating:
fhir: AllergyIntolerance.patient
[
... link <http://hl7.org/fhir/Patient/example>;
... Reference.reference [ fhir:value "Patient/example" ]
];
harold: I like having a type arc on a reference, which is specific about the target type, such as fhir:PatientReference instead of a generic fhir:Reference .
AGREED: If we include a type arc on a reference (such as the object of fhir:AllergyIntolerance.patient) it should be specific to the target type, such as fhir:PatientReference instead of fhir:Reference
dbooth: Still need to decide whether the type arc on a reference should be required, optional or not used.
ADJOURNED
This is scribe.perl Revision: 1.144 of Date: 2015/11/17 08:39:34 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) No ScribeNick specified. Guessing ScribeNick: dbooth Inferring Scribes: dbooth Present: Tony_Mallia Michael_Van_Der_Zel EricP Rob_Hausam David_Booth Thomas_Lukasik Harold_Solbrig Found Date: 26 Apr 2016 Guessing minutes URL: http://www.w3.org/2016/04/26-hcls-minutes.html People with action items: dbooth[End of scribe.perl diagnostic output]