See also: IRC log
harold: ont name had to be
different from the subject name. protege doesn't like something
being both a person and an ontology.
... Protege reasoner barely works with what we're doing.
Depends on order of imports -- very fragile.
SWITCHING TO SKYPE -- notify us here in IRC if you are trying to join!
harold: Protege insists on going to the web for imports, and its behavior depends on the sequence of buttons you push.
<scribe> ACTION: Harold to document the owl importing sequence needed to make it work, so that David can add it to the web pages [recorded in http://www.w3.org/2017/01/31-hcls-minutes.html#action01]
<trackbot> Created ACTION-75 - Document the owl importing sequence needed to make it work, so that david can add it to the web pages [on Harold Solbrig - due 2017-02-07].
harold: Barely handles bnodes, it
displays them correctly, but if anything goes wrong and you
look at individual nodes, protege complains of invalid names
for individuals.
... We can demonstrate this all works, but it is fragile. When
it doesn't work we don't know why.
... A simple soln to the owl:Ontology declaration problem might
be to declare another format, for example we could serve
fhir.owl to include that declaration (instead of
fhir.ttl).
... I plan to spend a little more time on this because we plan
to submit a paper on it.
dbooth: Would rather not add an fhir.owl format in addition to fhir.ttl unless someone shows a need for it.
harold: Editing fhir.ttl disk
file, there is a sequence of steps that works predictably, if
the order of imports is right. Disk file had owl:Ontology
statement AND two import statements, and that worked.
... But i wanted to move my import statements to my use case,
and that didn't work. Behavior depends on order of
imports.
... fhir.ttl from the FHIR site is OWL Full. Need to pare it
down to EL.
... We currently need everything except fhir:value,
fhir:nodeType and fhir:index and fhir:div to be declared as
objectProperties. But at the moment we don't use domains,
ranges or cardinalities.
... My goal is to start with the minimum and see how much of
that we can add and still make the reasoner work.
dbooth: Please call yours fhirmin.ttl just to avoid confusion
harold: if in diagexample.ttl I
include an ont decl, and import fhirmin.ttl then it shouldn't
matter where you import fhirmin.ttl, but it does matter.
... If I remove the import statement from diagexample.ttl then
i get different results.
dbooth: If there is a predictable sequence that makes it work then let's start with that.
harold: Yes, I have one, but it also depends on the timing of button pushing.
protege supposedly is doing better on serialization: https://lists.w3.org/Archives/Public/semantic-web/2017Jan/0099.html
https://lists.w3.org/Archives/Public/semantic-web/2017Jan/0099.html
<scribe> ACTION: Eric to ask protege folks to have a hacking session around AMIA timeframe when Harold and Eric are in CA [recorded in http://www.w3.org/2017/01/31-hcls-minutes.html#action02]
<trackbot> Created ACTION-76 - Ask protege folks to have a hacking session around amia timeframe when harold and eric are in ca [on Eric Prud'hommeaux - due 2017-02-07].
eric: I made FDA data work -- downloadable, it worked.
dbooth: Time to update the examples?
eric: Need to come up with URLs for the owl:Ontology statement
harold: protege does not want the subject and ont to be the same URL. But we could add .owl onto the end.
dbooth: Do we know that a FHIR resource URL must never end with a / or a fragID?
eric: i think so, if we restrict ourselves to the LD set.
<scribe> ACTION: Harold to find the FHIR resource URI regex [recorded in http://www.w3.org/2017/01/31-hcls-minutes.html#action03]
<trackbot> Created ACTION-77 - Find the fhir resource uri regex [on Harold Solbrig - due 2017-02-07].
http://dbooth.org/2017/ld.html
please take a look and give me any feedback/suggestions
ADJOURNED
This is scribe.perl Revision: 1.148 of Date: 2016/10/11 12:55:14 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 Present: EricP David_Booth Harold_Solbrig Found Date: 31 Jan 2017 Guessing minutes URL: http://www.w3.org/2017/01/31-hcls-minutes.html People with action items: eric harold[End of scribe.perl diagnostic output]