See also: IRC log
<trackbot> Date: 15 September 2015
dbooth: Tony unable to make it
today, and did not finish updating his ValueSet doc.
... No quorum. Call officially canceled, but those of us here
can hang out and talk if desired.
eric: Does HL7 care about ShEx?
lloyd: hl7 doesn't know about
ShEx. The portion of HL7 that knows/cares about RDF is sub 5%.
Of those, the ones who expect to do complex validation is even
fewer, in part because it has never been possible before.
... If ShEx were to do what we want it to do, would the actual
tools that get built be able to handle out volume and process
in a timely manner and provide back human-intelligible feedback
messages sufficient to use as an alternative to what we have
now.
eric: i was thinking of it not as alternative, but something for the RDF plane rather than XML/JSON plane.
lloyd: Presumably we could design our RDF to enable easy validation, but that's not the main reason for RDF. The main reason is reasoning.
eric: My intent is for it to have similar expressivity as relax-ng.
lloyd: slicing, where type is inferred from the name ...
eric: ShEx may be an advantage
for that.
... Claude was excited about expressing profiles in ShEx.
lloyd: I'm happy for it to happen. I think the percent who will touch it would be less than 5% -- RDF in general.
eric: Goal is to make it so that those 5% have access to validation tools.
lloyd: If it works it would be
cool, but would not change my world. I'm happy for the RDF.
I've been excited about RDF before, but from a production
perspective we're a ways away from that. I'm being
cautious.
... If ShEx met all of our requirements, would it be useful in
real world production env? If so, we can spend.
eric: Complexity is lower than
SPARQL. Triplestores are behind XML stores, so RDF platform is
at a disadvantage.
... But I don't think ShEx would be less performant than XML
validation.
... So if you can buy a bigger computer for your triplestore,
what would you get out of having a validator in the pipeline?
Obvious reason is to have less garbage coming in or going out.
But ShEx translation abiliity is also interesting.
... E.g., mapping between FHIR and CIMI models, which Claude
and I worked on.
... In principle could project queries through also.
<ericP> http://shexspec.github.io/extensions/Map/
<ericP> http://shexspec.github.io/extensions/Map/#bpexample
lloyd: if you think there will be real world use, I'm happy to put some energy into it.
<Zakim> dbooth, you wanted to say I expect us to use ShEx as reference implementation for FHIR RDF-to-XML translation.
eric: SHACL group is producing (effectively) OWL, so whenever you have sliding, you end up with something awkward. TopQuadrant is advocating SPIN. Ended up with them proposing a spec that is defined by their implementation.
<ericP> www.w3.org/2015/Talks/0909-shex-egp/#(4)
eric: In SHACL, treatment of more than one property is entirely different from treatment of one property.
dbooth: Possibility of connecting SHACL and ShEx?
eric: we would have to write down
a ShEx document in RDF, have it deal well with slicing and
repeated properties.
... If we did that, and were forceful, we could 85% chance get
our way.
lloyd: Forceful?
eric: Forceful means joining the
group, saying "here are our use cases", and saying "we want
this behavior".
... It will cost 20 hours of conference call time, and swap-in
time, to argue for things.
... If we don't do anything, and have no intersection between
things and have to write it completely differently if you have
one property vs two, then you end up with something that is
expressible in SHACL.
... That would be unfortunate.
dbooth: Can we really pull these two user communities together, ShEx and SHACL?
eric: Dublin Core people want to
write things in ShEx style. They've found OWL-like style didn't
meet their needs.
... My guess is that we can get the OWL heads to use ShEx, and
we'll get a larger community.
dbooth: What about the other way around? Would the SHACL folks be able to pull ShEx folks to their way of thinking?
eric: Probably not.
dbooth: I'm skeptical that it's worth trying to pull those two communities together.
lloyd: Seems useful to pull those communities together. Not sure I'm the right person to put that effort in.
dbooth: I suspect it is better for ShEx community to go forward with running code, and prove superiority through working code and use.
lloyd: my perspective: not sure
who in HL7 has the knowledge of the two specs and the HL7
requirements and the bandwidth to step forward as a
representative.
... Eric is best positioned to engage in this discussion. We
could support Eric. If it is feasible for Eric to continue, and
have HL7 credentials to add to eric's name to speak on our
behalf.
... Not sure the community would accept those additional
credentials.
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) Succeeded: s|www.w3.org/2015/Talks/0519-shacl-egp/#(4)|| No ScribeNick specified. Guessing ScribeNick: dbooth Inferring Scribes: dbooth Present: EricP David_Booth Lloyd_McKenzie Rob_Hausam Paul_Knapp_(last_half) Bill_Kleinebecker_(muted?) Found Date: 15 Sep 2015 Guessing minutes URL: http://www.w3.org/2015/09/15-hcls-minutes.html People with action items:[End of scribe.perl diagnostic output]