FHIR RDF usage

harold: We've started using FHIR/RDF and run into a few network issues.
... who what where when why ontology
... (namespace and maintenance)
... OWL code systems
... URI for code data type
... Conneg issues
... We'll need two formats of code systems: just RDF, and the OWL version.
... On Friday I'll show how we can use SNOMED with FHIR RDF.

dbooth: An example of the difference between the plain RDF and the OWL code systems would help.

harold: The basic RDF indicates the relationships between the codes. The OWL one says that these things are classes.
... We might use RDF/XML as the type for the OWL

dbooth: I'm uncomfortable with that approach. Needs more discussion when we have time.

harold: Rules for forming URIs for code systems are needed.
... The URI for a LOINC code for example is not in the FHIR JSON.
... We're using ShEx as a schema for the RDF.
... Another issue: fhir.ttl (previously called the FHIR ontology, but I'm now calling it the FHIR metadata catalog).
... As long as people realize that it isn't about people, but about records about people, then it's okay.
... At the moment it is auto-generated from the structure def.
... The stuff isn't all in there.
... Mappings, datetime ... there are 3 datatypes that OWL does not recognize, so there are some edits we need to do to use it with OWL.
... xsd:date and xsd:time are not in OWL. So we change them to xsd:string or xsd:datetime to make them work.
... Another issue: valuesets. Add them to fhir.ttl
... We don't have a good set of URIs for marking up healthcare, except perhaps fhir.ttl. Somebody could use it to mark up all sorts of other documents.
... Starting to use it in I2B2. FHIR concept codes.

lloyd: Since this is still in flux, should we be publishing URIs for them? They may change.

grahame: Some of the turtle generation is version specific and some isn't.

dbooth: RDF experience is that it's best to guess on permanent URIs

eric: And the process says that they may change until Candidate Recommendation

grahame: We have finer distinctions about stability in the FHIR publication process.

dbooth: But it would not be helpful to give temporary URIs that are different than permanent ones.

harold: Might be able to finalize Patient and Observation before others.
... Somebody needs to be the RDF implementation into HAPI

grahame: James would have to do a HAPI serializer.
... I've done most of the work for it. If we share the code base it would be a huge leg up.
... I add jena to the build tool to use sparql, but it adds a lot of code.

michael: Ewout wants the C# implementation to make the RDF optional, and that seems feasible.

dbooth: Might take the same approach for HAPI server.

harold: Another issue: Might want to publish fhir.schema.org
... FhirPath ?

grahame: We'd have to do more work to make FhirPath work directly against FHIR/RDF

eric: FhirPath is basically a subset of sparql paths.
... FHIR slices would take more thinking or giving up. :)

grahame: predicates are important in a schema language.

eric: In shex we have an extension mechanism that can be used for that. Also discussed creating an extension based on sparql to capture what you need.
... What is the cleanest way to represent the access to a property path and comparisons to other things you have accessed?

harold: one goal in shex was human readability. if you embed sparql then it is much less comprehensible.

dbooth: followups?

grahame: reworking W5. I need to do that right away.

harold: Also need to turn the fhir metadata vocab back into the fihr ontology.

lloyd: When we put w5 into logical model, does it allow us to represent event patterns also?

grahame: Not currently in the generated turtle.

lloyd: Right now the sources for the pattern mapping .. the linkage to the pattern .. is not there.

grahame: it will be.

ADJOURNED

This is scribe.perl Revision: 1.152  of Date: 2017/02/06 11:04:15  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00)

Succeeded: s/URI/... URI/
Present: David_Booth Harold_Solbrig Grahame_Grieve Josh_Mandel Michael_Van_Der_Zel EricP LLoyd_McKenzie several_others
No ScribeNick specified.  Guessing ScribeNick: dbooth
Inferring Scribes: dbooth

WARNING: No date found!  Assuming today.  (Hint: Specify
the W3C IRC log URL, and the date will be determined from that.)
Or specify the date like this:
<dbooth> Date: 12 Sep 2002

People with action items: 

WARNING: IRC log location not specified!  (You can ignore this 
warning if you do not want the generated minutes to contain 
a link to the original IRC log.)