Meeting minutes
<AndyS> https://
<Zakim> pfps, you wanted to comment on unstar
<pfps> It's unclear to me what role unstar now plays and what "two things" it relates together.
<Zakim> enrico, you wanted to comment on subj position and unstar
<AndyS> AndyS: IIRC we are having triple terms in the subject position in the semantics; in the RDF abstract data model (and hence syntaxes) it would be the more restrictive having triple terms only in the object position.
<pfps> There are lots of properties that mappings that do not preserve semantics can have. One of them is some sort of encoding. One desirable property of an encoding is some sort of reversability.
<Zakim> niklasl, you wanted to comment on unstar information preserving
<niklasl> w3c/
<gb> Issue 49 Define an interpretation of Triple Terms (by niklasl) [needs discussion]
<Souri> Just to be clear regarding triple-term position related decision: Has it been decided if N-Triple for RDF1.2 would allow having triple-term in the subject position?
<Zakim> pchampin, you wanted to propose an alternative definition of information preservation
<AndyS> To Souri - "decided" is too strong (we have not "resolved") but otherwise, from my "IIRC", NT would not have triple-terms in the subject position.
<Zakim> AndyS, you wanted to ask about where we are re: graph vs dataset (or both)
<thms> To Souri: IIRC the poll was about the semantics, but not about the possibility of triple terms in subject position (and neither about some syntaxes)
<pchampin> I was convince by niklasl's argument, as reported by gkellogg
<AndyS> We are looking at the "graph" flavor of unstar. (I will note that on the issue/PR for the wider audience)
<Zakim> enrico, you wanted to comment on strongly disagreeing unstar being only about syntax
<doerthe> I wonder wether the injectivity of our interpretation function could cause problems, but curious to see the bijection
<Zakim> enrico, you wanted to comment on not reusing the rdf old reification vocabulary
<niklasl> I lean towards enrico's point about using dedicated terms for unstarred triple terms
<niklasl> Explicit example of how I see it: https://
<doerthe> OK, maybe that was my point, so what is the bijection then?
<william_vw> unicity = uniqueness?
<niklasl> Unless you name a graph with that name?
<AndyS> and for graph-style use of the unstar vocabulary?
<thms> w3c/
<gb> Issue 114 Un-star operation to support RDF Dataset Canonicalization? (by niklasl) [needs discussion]
<thms> # a mixed environment
<thms> :r1 rdf:reifies
<thms> <<( :s :p :o1 )>> ,
<thms> <<( :s :p :o2 )>> ,
<thms> [ a rdf:Triple; rdf:subject :s ; rdf:predicate :p ; rdf:object :o3 ] ,
<thms> [ a rdf:Triple; rdf:subject :s ; rdf:predicate :p ; rdf:object :o4 ] ,
<thms> :g1 .
<thms> :g1 a rdf:Graph .
<thms> :g1 {
<thms> :s :p :o5 ,
<thms> :o6
<thms> }
<niklasl> rdf:TripleTerm owl:hasKey (rdf:tripleTermSubject rdf:tripleTermPredicate rdf:tripleTermObject) .
<niklasl> That would fix it?
<pchampin> sameAs is defined on OWL; I think it should be possible to express the axioms on owl that allow us to infer that the two nodes of type TripleTerm are indeed sameAs, no?
<pchampin> using keys?
<niklasl> Yes, I just wrote that. :)
<enrico> r1 rdf:reifies <a b c>. r2 rdf:reifies <d e f>.
<pchampin> thms: for me, the problem is that if you unstar, and then "restar" (inverse of unstar) your graph, it may not round-trip
<enrico> r1 rdf:reifies _:b1. _:b1 :subject a. _:b1 :predicate b. _:b1 :object c.
<enrico> r2 rdf:reifies _:b2. _:b2 :subject d. _:b2 :predicate e. _:b2 :object f.
<AndyS> r1 rdf:reifies <<(:a :b :c)>>. r2 rdf:reifies <<(:d :e :f )>>.
<enrico> a owl:sameAs d. b owl:sameAs e. c owl:sameAs f.
<enrico> SHOULD ENTAIL
<enrico> _:b1 owl:sameAs _:b2.
<doerthe> I still dislike the injectivity (but can live with that after the discussion with pfps :) )
<thms> could we say that an rdf-star reifier that rdf:reifies only one triple term can be represented as RDF standard reification?
<niklasl> thms: I have en example of that (using OWL)....
<thms> niklasl do you happen to have a link?
<niklasl> It's not easy on the eyes, but: https://
<thms> niklasl :) thanks!
<niklasl> Caveat: not updated to the correct unstarred forms
<doerthe> and for that I tried to make my argument short :(
<Zakim> AndyS, you wanted to ask about whether old-style reification to RDF-star is in fact a different algorithm
<pchampin> +1 to AndyS, because most old style reifications out there are probably more reifiers than triple terms
<thms> _:r a rdf:Statement ; rdf:reifies <<( :s :p :o )>>
<AndyS> We also have more choice/control of the property names in unstar.
<doerthe> mmm, following your example enrico, I can see that it comes in hand that if I have
<doerthe> r1 rdf:reifies _:b1. _:b1 :subject a. _:b1 :predicate b. _:b1 :object c.
<doerthe> r2 rdf:reifies _:b2. _:b2 :subject d. _:b2 :predicate e. _:b2 :object f.
<doerthe> _:b1 owl:sameAs _:b2:
<doerthe> here, the injectivity comes in handy. We then have some kind of functional :subejct, :predicate, :object and we are out of our multiple object problem.
<doerthe> I know that this is a counterargument to my position :)
<thms> _:oldstyleReifier rdf:subject [ rdf:subjectOf <<( :s :p :o )>>; ..
<doerthe> I still dislike, yet, I see your point, but I think the triple term could be anything
<thms> i mean this
<thms> _:oldstyleReifier
<thms> a rdf:Statement ;
<thms> rdf:subject [ rdf:subjectOfTripleTerm <<( :s :p :o )>> ] ;
<thms> rdf:predicate [ rdf:predicateOfTripleTerm <<( :s :p :o )>> ] ;
<thms> rdf:object [ rdf:objectOfTripleTerm <<( :s :p :o )>> ] .
<Zakim> niklasl, you wanted to comment on OWL...
<enrico> we have been kicked out!
<Souri> re-join?
<pfps> It's a sign!
<AndyS> AI
<doerthe> OK, I think I dislike the whole injectivity because that already has old rdf reification in mind, we see a syntactical triple structure while we could also say that the triple term denotes some resource