Semantic Web Health Care and Life Sciences Interest Group Teleconference

21 Jul 2015

See also: IRC log


David_Booth, Tony_Mallia, Bill_Kleinebecker, Brian_Pech, EricP, Lloyd_McKenzie, Rob_Hausam, Paul_Knapp_(briefly)
David Booth


<trackbot> Date: 21 July 2015

<scribe> scribenick: ericP

Term Info

<dbooth> eric: There's the problem of things that can be said in the info model, and you can go back and forth. And there's the notion that an uniform representation would allow you to coerce both of those into the same model, and then it woulwn't matter which way it came in.

<dbooth> ... If i parse it from a microparse model from snomed, alll the concepts have representations as act relationships and that allows you to populate an info model that looks like what you would have got if you had gottten all atomic snomed terms.

<dbooth> ... But i never actually wrote the microparser.

<dbooth> ... Anyone else working on that?

<dbooth> Rob: Don't know that anyone has done that.

<dbooth> ... Not seen anyone do the transform between RIM models and back and forth, but I'm hoping that we can do that with the RDF work that we're doing.

<dbooth> eric: it would allow something to say "invoke this microparser"

<dbooth> rob: how close to equivalence is needed?

<dbooth> eric: would like to see cases where it works, where it doesn't, and then understand what nuance and delicacy is needed for this.

<dbooth> rob: I'm very interested in it.

<dbooth> ACTION: EricP to create a microparser as a tool that takes an RDF graph and a map from SNOMED relationships to ACT relationship type codes and emits an RDF graph with only atomic SNOMED terms, due 21-aug-2015 [recorded in http://www.w3.org/2015/07/21-hcls-minutes.html#action01]

<trackbot> Created ACTION-29 - Create a microparser as a tool that takes an rdf graph and a map from snomed relationships to act relationship type codes and emits an rdf graph with only atomic snomed terms, due 21-aug-2015 [on Eric Prud'hommeaux - due 2015-07-28].


dbooth: no task force meeting last week.

CTS Ontology Workshop, Sept 23-25, Charleston SC

dbooth: previous week was about CodableThingy
... folks (tmallia) check stuff into github

tmallia: still working on Side-by-side
... haven't started working on the github artifact, which i assume is the mapping
... i need to get further in our disucssions.

dbooth: Charleston meetings are 23-25Sep
... with Barry Smith et al
... it would be good to have a couple weeks lead time so they can look at it
... it would be good to have a couple weeks lead time so they can look at it

tmallia: what's the audience?

<dbooth> link to charleston meeting: http://ncorwiki.buffalo.edu/index.php/CTS_Ontology_Workshop_2015

tmallia: i've been assuming knowledge of RDF and OWL, so I can write stuff in Turtle

Side-by-side example of two FHIR RDF approaches (Tony Mallia)

<dbooth> http://wiki.hl7.org/index.php?title=File:FHIR_RDF_Sample_side_by_side_comparisons.pdf

<dbooth> http://wiki.hl7.org/images/2/25/FHIR_RDF_Sample_side_by_side_comparisons.pdf

ericP: note that the wiki attachement page apparently displays the first comment which makes it appear to be V5.

tmallia: we need to settle on names so we can make the changes

<dbooth> line 89 and onward

tmallia: FHIR Code and CodableConcept are not direct translations
... I've created "locally-defined" classes, i.e. CodingThingy.{System,Version,Code,Display,Primary}

[discussion of capitolization leading to some model changes]

<dbooth> https://hl7-fhir.github.io/datatypes.html#coding

<dbooth> changing fhir:CodingThingy.Code to fhir:CodeThingy

tmallia: we're changing e.g. "fhir:CodingThingy.code [ a fhir:CodingThingy.Code ; ... ]" to "fhir:CodingThingy.code [ a fhir:CodeThingy ; ... ]"

<dbooth> changing the names of the green classes, lines 98-101

tmallia: object properties CodingThingy.{system,version,code,display,primary} remain the same
... [note capitalization]

<dbooth> scribe: lloyd

<dbooth> tony: CodeableConcept, Coding and Code will be subclasses of CodeableThingy, and those subclasses will be used to tag the RDF to indicate the source construct used.

<dbooth> eric: every FHIR property has a non-union range

<dbooth> tony continues on lines 199 and on

<ericP> Lloyd: where we [would] have untions, we have distinct properties

Object properties all have non-union types, so no need to declare type - implicit from property name.

Eric: do we want qualifiedCardinality or just cardinality?

Tony: issue w/ using Protege editor

<dbooth> david: Not sure we'll want all those restrictions on the three CodeableThingy subclasses, (such as maxQualifiedCardinality 0 for version), because when data is merged and embellished, there may be more info added.

<dbooth> lloyd: if you have a Coding, then the system will be fixed and behind the scenes.

<dbooth> ... Different instances of Coding can have the same code and system, but different displays. Code never has a display.

<dbooth> ... You cannot have a Code with an unknown system. A code must always be tied to a valueset with a fixed binding.

<dbooth> eric: In ShEx, we might say there is no system. That means that during transmission, it will cause an error if it is there.

<dbooth> ... But in OWL we can say the cardinality is 1 even though the expressed data does not show one.

<dbooth> ... We just haven't seen it.

<dbooth> lloyd: For a Code, the system MUST be known. But for CodingThingy in general, you do not necessarily know.

<dbooth> eric: We're writing OWL, which is open world, but we can extract from that what we want to use for closed world validation.

<dbooth> lloyd: If we say that system exists (for Code), it will not exist in the instance data, so there would be a problem.

<dbooth> eric: We might say "dont' serialize this, though it exists"

<dbooth> ... There's a system for Code, but don't write it.

<dbooth> david: Two different use cases for the OWL: describing the RDF that exists in the transmitted instance data, versus describing the information that is known (but not necessarily transmitted in the instance data).

<dbooth> eric: I think the latter

<dbooth> dbooth: I think the former.

<dbooth> eric: I think we'd do the former with ShEx.

<dbooth> ... Could have a tool that reads the OWL and spits out the ShEx.

<dbooth> ... Could add annotation properties that aid the translation to ShEx.

<dbooth> david: I think we should avoid having some RDF that initially comes in as Code, with no display property, and then a display property gets added, and suddenly it is no longer a Code.

<dbooth> tony: starting on lines 254, we need to decide on a proper name for CodeableThingy. I've proposed something.

<dbooth> eric: Concern about fhir:Concept is that everything is a concept.

<dbooth> david: it's a superclass of CodeableConcept.

<dbooth> lloyd: Suggest CodeableThing and CodingThing

<dbooth> david: Or SuperCodeableConcept

<dbooth> david: CodeableSuperclass

<dbooth> We will continue to collect and debate naming ideas.

Summary of Action Items

[NEW] ACTION: EricP to create a microparser as a tool that takes an RDF graph and a map from SNOMED relationships to ACT relationship type codes and emits an RDF graph with only atomic SNOMED terms, due 21-aug-2015 [recorded in http://www.w3.org/2015/07/21-hcls-minutes.html#action01]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.140 (CVS log)
$Date: 2015/07/21 16:29:38 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
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/tmalia/tmallia/g
Found ScribeNick: ericP
Found Scribe: lloyd
Inferring ScribeNick: Lloyd
ScribeNicks: ericP, Lloyd
Present: David_Booth Tony_Mallia Bill_Kleinebecker Brian_Pech EricP Lloyd_McKenzie Rob_Hausam Paul_Knapp_(briefly)
Found Date: 21 Jul 2015
Guessing minutes URL: http://www.w3.org/2015/07/21-hcls-minutes.html
People with action items: ericp

[End of scribe.perl diagnostic output]