<trackbot> Date: 26 May 2015

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

<dbooth> Scribe: Marc_Twagirumukiza

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

Tony: We had been working on this also in small group
... there are line number which can help in the document
... section 2.1 (see L25-28)
... saw that the github version changed
... I had originally done this, the value has no specific domain
... FHIR primitives

<Rafael> Rafael here

<dbooth> dbooth: Line 25 .coding is supposed to be a predicate URI TBD?

Tony: section 1.2 and sub-section
... The original github was wrong... it has been corrected

ericP: That means the algorythm was wrong?

Tony: section 1.2.2 shows it
... the FHIR sting has a value which is a FHIR primitive

<dbooth> Tony: coding.code.value needs to be an object property, not a datatype property, for extensibility -- to allow other properties to be attached.

Tony: section 2 line 79
... and following
... there is a problem of blank nodes, some inferences work some others not
... this the first step L130

Lloyd: can you explain more--- don't understand well

Tony: L115 un-labelled blank node and L130 labelled blank node

Lloyd: I can have 2 different codings

Tony: they can have different declarations

<dbooth> Tony: the same singleton URI can have different declarations in different places

dbooth: It would help if we differentiate things here

ericP: if we have a single term and we need to have extension we would not go to the single term

<dbooth> eric: If we attach extra properties to a singleton URI at all, we're saying that they are true everywhere. We'll instead want a node that is *related* to that singleton, so that we can attached extensions to that intermediate node.

<dbooth> lloyd: We can have 5,000 instances, each referring to the same code, but each one must independent.

<dbooth> tony: why?

Lloyd: ech term has to be completery different from each other

Tony: why they need to be separated?

Lloyd: I think I agree with you

<dbooth> lloyd: why do we need a singleton at all? why not have a class?

<dbooth> ... we could have an inference rule that sees that class.

Tony: L131
... blank node could be a unique URI

<ericP> http://piratepad.net/hcls2

Tony: the nice thing here the label is attached to the name of the class and the label is propagated to the instances
... in SNOMED f.eg is attached to the Class

If you change the name of L135 the label will go

Tony: If you don't declare the individual,.... I dont think you can infer the label

Lloyd: we need to investigate what cause this

<dbooth> dbooth: Would help to separate out the question of punning. It's a convenience that can be used, but doesn't do anything you could not do with separate URIs.

Tony: property chain seems to work with the reasoner

dbooth: ... named individuals Vs blank node
... we have to figure out if we need unique IRI

correct not IRI but URI

dbooth: How are we going to proceed?

ericP: we need to figure out if we need to something there

dbooth: we need exemples

<ericP> example of blank nodeode differentiation

Tony: Lets' use the exemples we have there
... We can start with an assumption that the blank node works

ericP: Using the exemples I had before

<dbooth> ACTION: David to ask Michel Dumontier about blank node inference problem [recorded in http://www.w3.org/2015/05/26-hcls-minutes.html#action01]

<trackbot> 'David' is an ambiguous username. Please try a different identifier, such as family name or username (e.g., dbooth, dderour, dhansen2, dnewman, dshotton, dwild).

dbooth: to create an action

Tony: I don't see the direction, I would take it back to blank node
... Concept Binding Internal-- Section 3

dbooth: I don't see the class at L294
... I was expecting allergy confirmed there L294

Tony: see line 300 is there
... there is a Class

Lloyd: L299 has to be URI

Tony: am in favour not to use the reasoner for that, it's not scalable

-

<dbooth> Scribe: dbooth

eric: tony wants a single pass; lloyd wants a clear def of what the RDF is. We could say that the transformation does two things.

lloyd: My understanding of the requirements is that we will not inject any information other than what is in the instance data. Nothing else can use it in round tripping.

tony: It's part of the valueset definition.

lloyd: valueset definitions don't give you URLs for SNOMED concepts.
... There are no SNOMED concept URIs in the def.
... When you convert FHIR XML to RDF, you cannot introduce the SNOMED URIs.
... Cannot add info except what's in the metadata that is part of the resource definition.
... The resource defn knows cardinality, etc. but not about SNOMED URIs.
... Therefore the SNOMED URIs cannot be a part of the round tripping process.
... But it *can* be introduced later by an ontology, by SPARQL rules, by many other ways.
... The reference transformations between FHIR XML and FHIR RDF can only use the FHIR XML instance data and resource definitions, but nothing else.
... We might make an exception for FHIR-maintained code systems, but cannot do it with anything else, because there are so many of them and they're always changing.

tony: No standard way to form vocab URIs from the combination of Coding.system URI and Coding.code .

ADJOURNED

<ericP> http://piratepad.net/hcls2

s|s/BN/blank node/g||

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/I had/Tony: I had/
Succeeded: s/BN/bnodes/
Succeeded: s/BN/bnode/g
FAILED: s|s/BN/bnode/g||
Succeeded: s/bnode/blank node/g
Found Scribe: Marc_Twagirumukiza
Inferring ScribeNick: Marc_Twagirumukiza
Found Scribe: dbooth
Inferring ScribeNick: dbooth
Scribes: Marc_Twagirumukiza, dbooth
ScribeNicks: Marc_Twagirumukiza, dbooth

WARNING: Dash separator lines found.  If you intended them to mark
the start of a new topic, you need the -dashTopics option.
For example:
        <Philippe> ---
        <Philippe> Review of Action Items

Default Present: DBooth, ericP, +1.801.949.aaaa, rhausam, lloyd, +1.978.794.aabb, Tony, Marc_Twagirumukiza, Rafael, [IPcaller]
Present: David_Booth EricP Rob_Hausam Lloyd_McKenzie Tony_Mallia Rafael_Richards Marc_Twagirumukiza
Found Date: 26 May 2015
Guessing minutes URL: http://www.w3.org/2015/05/26-hcls-minutes.html
People with action items: david