See also: IRC log
<trackbot> Date: 29 April 2015
issues list
blank node
side-by-side comparison: references
http://www.w3.org/2014/HCLS/track/issues
http://www.w3.org/2014/HCLS/track/issues
http://www.w3.org/2014/HCLS/track/issues
http://www.w3.org/2015/04/28-hcls-minutes.html#item06
"RESOLUTION: agreed that we should represent Observation.code as an instance of one or more classes"
<ericP> hcls pirate pad
tony: loading into protege,
looked at what the reasoner could determine.
... I found that if the person has a URI-named person with hair
color "brown" then the reasoner can figure out that the person
is a member of BrownHaired class, but if it is blank node then
the reasoner does not make that conclusion.
<Zakim> dbooth, you wanted to ask if this is a fundamental limitation of using a blank node in this position or just a limitation of this particular reasoner
claude: dont' know, but I know that certain reasoners have limitations.
tony: I tested hermit and fact++
and snorocket
... my conclusion: blank nodes are fundamentally dangerous.
eric: with SPARQL, a query finds both of Frank's sons. But a DL query does not, just as tony's experiiment showed.
Options: 1. use blank nodes and let applications deal with the problem by minting skolem URIs or something else; 2. define a standard way to form a URI; 3. use a skolem URI using the well-known convention defined by RDF.
http://www.w3.org/TR/rdf11-concepts/#h3_section-skolemization
claude: regarding #2, if one of the parents has a unique URI, would that be an option?
dbooth: Yes, if that node is
uniquely identifiable via a path to it.
... Usually that is done with inverse functional
properties.
eric: I did something like that
when I converted RIM, and those URIs are repeatable becuase
they're based on the XML. But the challenge is that if it is
done by position -- "the 4th act relatinoship" -- then that
might not reflect the real semantics of what should identify
that thing.
... if I get the same XML FHIR document with something
additional inserted, then the URI would change, which means
other references to it would change.
... But if it is done with something more stable, like date
perhaps, then the URI would be stable if the data is
regenerated.
claude: If there's a way to do that, it may be desirable.
tony: that would improve the
robustness, because we dont' know what tools will be
used.
... It's probably okay if Coding and Code are left as blank
nodes because you never need to navigate to them, because the
info has already been pushed up into the class.
... But CodeableConcept needs to be a URI.
... But we also have cases where Coding and Code are by
themselves and would need to be URIs.
claude: Code is a degenerate case of Coding (where the namespace is fixed). But CodeableConcept is different because it is a concept whose meaning is represented by the collection of Codings that it contains, which must be synonyms.
rob: There's no way that they will be identical.
dbooth: the CodeableConcept instance must be a member of the intersection of the classes.
Options: 1. use blank nodes and let applications deal with the problem by minting skolem URIs or something else; 2. define a standard way to form a URI; 3. use a skolem URI using the well-known convention defined by RDF.\
eric: Use cases should determine which blank nodes to address with option 2 or 3.
claude: Prefer option 2.
2. define a standard way to deterministically form a URI based on the FHIR XML
scott: #2 places a higher burden on us.
tony: Also, #2 will work with prefixes.
eric: I would expect it to be
relative to the base URI.
... the CCDA work i did reflected the XML hierarchy in
generating the URI.
<scribe> ACTION: Eric to show an example of generating a URI from the XML structure [recorded in http://www.w3.org/2015/04/29-hcls-minutes.html#action01]
<trackbot> Created ACTION-22 - Show an example of generating a uri from the xml structure [on Eric Prud'hommeaux - due 2015-05-06].
<ericP> http://www.w3.org/2013/C-CDA/IJ.xml
eric: ... the xpath
claude: if you do this as a
CodeableConcept then it is only relevant to what it is attached
to.
... If we do a path relative to the containing resource, then
if I have the exact same concept defined the same way
elsewhere, then I cannot say that they are the same entity.
http://www.w3.org/2013/C-CDA/IJ.xml
<ericP> structuredBody
<ericP> _:section0_allergyProblemAct0_entryRelationship0
<scribe> ACTION: Tony and Eric to propose an algorithm for generating URIs for CodeableConcept instances [recorded in http://www.w3.org/2015/04/29-hcls-minutes.html#action02]
<trackbot> Error finding 'Tony'. You can review and register nicknames at <http://www.w3.org/2014/HCLS/track/users>.
ADJOURNED
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) No ScribeNick specified. Guessing ScribeNick: dbooth Inferring Scribes: dbooth Default Present: ericP, Tony, DBooth, rhausam, claude, Marc_Twagirumukiza, +31.62.427.aaaa, mscottm Present: ericP Tony DBooth rhausam claude Marc_Twagirumukiza mscottm Found Date: 29 Apr 2015 Guessing minutes URL: http://www.w3.org/2015/04/29-hcls-minutes.html People with action items: eric tony[End of scribe.perl diagnostic output]