Meeting minutes
R6 prep
w3c/
dbooth: Kindling PRs have not been merged yet. Others have.
tim: I'll message Grahame.
dbooth: treeRoot change from class to individual does not change the data on the wire
… I'll plan to improve the wording later: ""For purposes of validating FHIR RDF data, the logical meaning of the OWL ontology is normative. However, ..."
tim: Need to clarify that what appears in the instance data is what cannot change without being a breaking change.
dbooth: Anything else urgent fo rR6 deadline?
jim: I have not had time for #170 semantics of hash IDs
Sorting the ontology
tim: Pretty straight forward. I did it using an open source lib. Not sure if the would allow it.
… Apache jena uses random order.
https://
tim: This one sorts: atextor/
dbooth: Would it be kinder to wait until after the R6 deadline to do this one?
ACTION: dbooth to add this to our R6 checklist
ACTION: dbooth to reference the PR from the jira ticket
Example files with empty blank nodes #206
jim: I'll look at it
THO LOINC.html page shows wrong IRI stem #196
gaurav: I'll work on it.
Update FHIR RDF Playground to R6
dbooth: I think it's valuable for simple testing
tim: Agreed, but I'm focused on real application
… Higher priority would be validation tool that can be used in the real world.
Turtle examples of Bundle.entry.resource use a list but should not #194
jim: I'll look at it
ValueSet bindings for ontology and validation #167
tim: Would allow you to run a reasoner and get inconsistencies if the wrong code is used, i.e., closed world validation w OWL. But doesn't work w bindings to CodeableConcepts
(EricP joined)
tim: We can even generate an IRI for it, that would bring you to the valueset (almost).
ericp: We've been adding type arcs to codings. Should we to it by the literal code value or the IRI that we gen for the code?
tim: We would not get the closed-world validation using the IRI.
tim: you would have to manual expand sets of codes in advance
… And to use shex, if you want to expand the codes, you would have to dynamically expand the codes in shex too.
… If you have SNOMED in OWL, you could use OWL to do the expansion.
… Could enable a terminology service that would be OWL/shex native
… So if we set up OWL and shex this way, someone could set up a terminology service and use OWL and shex natively.
ericp: Anybody have a pet OWL implementation that would be easy to extend with user-defined functions, to expand a valueset dynamically?
tim: I did that using apache jena, and also w OWL API.
… Also was able to, in kindling, created the class axioms of the valuesets
dbooth: Lifecycle issue: lifecycle of terminologies is independent of FHIR version lifecycle.
tim: If the ont just points to the valueset IRI.
… The resources already have valueset bindings.
… Profiles have rules that can constrain valuesets
ken: But you can have bindings to multiple terminologies
tim: That's handled by what a valueset is
… I would also represent a CodeSystem as a class of codes
… Then a valueset would be a union of subsets of the CodeSystems
ken: can this prove that some valuesets are disjoint? e.g., status is used by multiple resources.
tim: Codes are identified by a combination of code and system, so they're disambiguated that way.
ken: Sometimes they use the same CodeSystem for "status", across different resources, like Observation status vs some other status.
dbooth: Does this depend on having IRI stems registered for the CodeSystem?
tim: Yes
ShEx generator
ericp: It's ready for PR now.
dbooth: Let's stay on the zoom to figure out how to do the PR
ADJOURNED