Meeting minutes
<niklasl> owl:sameAs a rdf:Property # so it may be in the intersection of ReificationProperty and "regular" properties...?
<niklasl> +1 for the Literal:ish view of triple terms
<niklasl> owl:sameAs is used with blank nodes a lot (OTOH it's also very misused)
<AndyS> owl:sameAs rdfs:domain owl:Thing .
<AndyS> owl:sameAs rdfs:range owl:Thing .
<niklasl> I could live with the stricter syntactic restriction (noting that I am also for only allowing rdf:reifies at this point / for RDF 1.2).
<Zakim> pchampin, you wanted to discuss the notion of "minimal" as argued by Souri
acl tallted
<TallTed> +1 SHACL. I meant to mention it.
<william_vw> pchampin sorry :_)
<AndyS> I very much hope that our non-REC guide to using RDF 1.2 reification covers good and bad patterns.
<TallTed> "warn when loading data that is inconsistent with previously loaded data" vs "warn when loading data that is internally inconsistent" vs "prevent loading data that is internally inconsistent" vs "let me load anything" (among other options)
<TallTed> such optional switches are (or should be) always viable. forcing any of these via spec seems not so viable.
<niklasl> +1 to AndyS on good/bad patterns (adding to my draft notes)
<william_vw> +1 to pchampin on having additional layers cope with this
<pchampin> I think that Souri's use-case is about one-to-many, not one-to-one
<pchampin> I don't see any reason to forbid/prevent subproperties of rdf:reifies (in RDFS, that is)
<niklasl> ... apart from semantics...
<gkellogg_> We have a syntax for arbitrary predicates used with Triple Terms in Turtle: `:r :p1 <<(...)>>>`
<pchampin> gkellogg_ +1
<niklasl> +1 to gkellogg_
<Souri2> I was thinking about a shortcut akin to :r | :s :p :o .
<niklasl> My point was a lot about the shorthands catering for the (so far) "understood/non-experimental" case.
<gkellogg_> Short cuts are to aide common usage patterns; trying to invent a short cut style that extends to arbitrary predicates would seem outside our scope.
<Souri2> :r [:p1] | :s :p :o .
<Souri2> or something along that line
<tl> gkellogg_ but that's what the annotation syntax does
<gkellogg_> (Best start adopting to the actual syntax: `:s :p :o ~ :r`)
<gkellogg_> `:s :p :o ~ :r` could result in a different predicate than `<<:s :p :o ~ :r>>`
<Souri2> :s :p :o ~ :r | :p1 . ?
May I say that I really really dislike the "~" syntax?
<william_vw> under datatype entailment: "RDF processors MAY treat an unsatisfiable graph as signaling an error condition, but this is not required."
<william_vw> https://
<Souri2> +1 to PA
niklasl
<gkellogg_> `<<(...)>> a rdf:TripleTerm`
<niklasl> That can be entailed, but I see no need to write it.
<niklasl> true a xsd:boolean .
<Souri2> disallowing triple-term in position other than the object position (as a syntactic restriction in RDF) => none of the properties in the rdf:ReificationProperty class can have an inverse property
<niklasl> ex:nameOf owl:inverseOf foaf:name .
<TallTed> there is no requirement that every property must be invertable
<Souri2> +1 to PA's point: "20"^^xsd:decimal :ageOf :john . is not allowed in RDF
<gkellogg_> `rdf:TripleTerm is rdf:type of <<(...)>>>`
<TallTed> sadly, I have to bounce (already 5 past official end time)
<TallTed> "is rdf:reifies of"
<niklasl> It's more of an "expressivity omission"...?
<niklasl> +1 to counter-examples