<trackbot> Date: 17 June 2015

Agenda

Eric: transformation software

Tony: References, in side-by-side

dbooth: ordered lists

Transformation Software

Eric: You give this an input XML doc, and it spits out the RDF

<ericP> FHIR/XML to RDF (Turtle)

<ericP> FHIR/XML to RDF (Turtle)

<ericP> FHIR/XML to RDF (Turtle)

eric: Put in a URL for some FHIR XML, and it gives you back FHIR RDF, in Turtle.
... The ShEx for this is generated from the FHIR schema definitions.
... Plus there's some special handling code.
... I'm also working on getting it to handle the profiles used for defining, e.g., specializations to quantity.

<ericP> parameters

dbooth: eric, please add a link to explanation of the parameters of that form
... suggest that we adopt eric's ShEx transformation as our first reference implementation for transforming FHIR XML to FHIR RDF
... FHIR XML and JSON use spec descriptions and reference implementations.
... Suggest we do the same for FHIR RDF.

tony: What about the ontology?

lloyd: That's separate.
... We're really defining the base conversion algorithm, for converstion from FHIR structure def and value set and possibly concept map resource instances to OWL and/or RDFS expression that means the same thing, in terms of defining the classes.

dbooth: 1. HTML page that describes (mostly in prose) the FHIR RDF (instance data).
... 2. Multiple reference implementations for generating FHIR RDF instance data (from FHIR XML, for example).

lloyd: Rather phrase that as adjusting existing implemntations to handle RDF also.

dbooth: 3. FHIR RDF Ontology. This will be generated (partially) from the FHIR conformance resources.

tony: is there a formal def of the transformation?

lloyd: no. we describe how to express instances from the structure model in particular serializations.
... 4. A page describing the OWL representation of FHIR artifacts. Describes #3. What is RDF/OWL, where might they be useful? HOw do core concepts in structure defs correspond to RDF constructs? How to integrate FHIR ontology w other ontologies? e.g., using SNOMED ont.
... Whole point of FHIR is to make it easy for implementers to do things. There will be people who come to FHIR who may want to know how to use FHIR RDF, and we need to give them that guidance.

eric: How does the ShEx transformation fit in?

lloyd: conformance resources: conformance, structure, value systems, concept maps (def links between terminologies or structures), data elements, operation defs, naming system (here's how different URIs map to each other -- not sure how to expose in OWL -- maps one namespace to another), test scripts.

<Zakim> dbooth, you wanted to suggest that the ShEx implementation be used as a ref impl for the RDF ref impls :)

dbooth: The ShEx script could be viewed as a formal def of our desired XML-RDF translation

lloyd: It
... It's a tech. I expect at least some of the ref imple to be modified to produce/consume RDF. The doc should be agnostic.
... Every ref impl will able to produce/consume RDF.
... Ref impl should be able to consume Turtle, RDF/XML and various RDF syntaxes. Libraries can convert between RDF implementations.
... That's on the instance data side.
... For class def side, that's not so clearly in the ref impl side, we should discuss whether the ref impl should produce/consume.

dbooth: What should they produce/consume for ontology?

lloyd: If an XML schema were provided, it could spit out an ontology for it.
... The pub process produces the ref impl, and they are used to produce the JSON versions. It also does a round trip check.
... Both for example instance data and for profiles.
... The pub process reads spreadsheets and produces profile XML files.
... The ref impl takes those and turns them into profile JSON files. Once we have RDF going, it will produce profile RDF files also.

dbooth: Seems like we need to better understand the pub process to know what parts to add in for RDF.

lloyd: Need to create the two HTML pages (equiv to existing XML and JSON syntax pages; and FHIR as ontology page, that talks about how FHIR structure defs and other conformance resource can be expressed using OWL)
... Need to have modified the supported ref impls to consume/produce RDF, just as for XML and JSON.
... Need to modify the build process, so when it produces JSON versions of everything and tests roundtripping, it also produces RDF versions of everything and tests roundtripping for that.

dbooth: suggest that the ShEx implementation be used as a ref impl for the RDF ref impls
... others agree?

tony: valuable to have it as an independent testing mechanism.

eric: allows people to validate in a generic tool without having to roll out a ref impl.
... Much use of the XML schema in the wild?

lloyd: yes. we use it internally to guide examples, and for validating.
... Some implementers use it and run into all of the issues associated with that.

AGREED: Use ShEx implementation as a reference implementation for the RDF reference implementations

eric: Could use some others hacking on the ShEx

dbooth: i am planning to do so, but have been trying to finish another project first.
... What needs to be done to move the ShEX to github for others to collaboration?

eric: Mostly people give me their github IDs, so that I can add them.
... today showed the XML-RDF direction. Still working on the ShEx part for the other direction: RDF->XML.

This is scribe.perl Revision: 1.140  of Date: 2014-11-06 18:16:30  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/softward/software/
Succeeded: s/schemas/conformance resources/
No ScribeNick specified.  Guessing ScribeNick: dbooth
Inferring Scribes: dbooth
Default Present: +1.310.266.aaaa
Present: Tony_Mallia David_Booth Lloyd_McKenzie EricP Don_Doherty
Found Date: 17 Jun 2015
Guessing minutes URL: http://www.w3.org/2015/06/17-hcls-minutes.html
People with action items: