Meeting minutes
<niklasl> I agree to suggest/advice but not commit.
<niklasl> As gkellogg has noted in the past, mapping LPG to RDF is similar to mapping JSON to RDF; there are lots of detailed choices, which (as AndyS just said) is context-sensitive.
<niklasl> I'd say the `<<(:a1 :TRANSACTION :a2)>>` is a reference the abstract relationship, and that is reified by the Transaction resource `:t1`. That is formally enough information; but users would *probably* prefer it to also be asserted. Ideally using the annotation syntax. (My too-late-for-this-meeting last email expanded on this.)
<niklasl> (my last email: https://
<Souri> Named-and-Asserted (atomic, not just shortcut)=> :e1 || :a1 :paid :a2 . Named-and-Unasserted=> :e2 | :a2 :paid :a3 .
<Zakim> niklasl, you wanted to say that *no* simple relationship is asserted in LPGs (only "occurrences")
<niklasl> So you don't want to change the definition of a graph being a set of triples? That is, it is enough with annotation sugar, but you want to expand it to something like `:a1 :paid :a2 . :e1 rdf:implies <<( :a1 :paid :a2 )>> .`?
<gkellogg> enrico: Many-to-one restrictions are implicit in the LPG transformation as described in my document.
<niklasl> the document is https://
<Souri> Unless we specifically use reification quads, every triple in RDF1.1 is asserted. RDF1.2 is enabling unasserted statements.
<niklasl> It's not hidden, it is transparent. It is not itself in the world as a simple relationship in this model. A "nicer" model would have it there.
<gkellogg> +1. An assertion implies that it is in the graph. A triple term, which is not in the graph itself is not an assertion. It may be a "claim" or "hypothesis".
<Souri> An RDF graph has this triple=> :s :p :o . This triple is asserted. Suppose I want to add annotations for this asserted triple. I need to add a "name" (not a reifier, because there is no reification happening here) to be able to say something about it.
<niklasl> Souri The graph is a set, so I don't know what "this" means, the membership of the abstract triple or "some occurrence"?
<niklasl> Technically this is the "frozen" version: https://
"final revised updated new 2024-06-23 version.txt"
"asserted triple" vs "unasserted triple". Calling "triples" "assertions" is better than calling "triples" "facts". "Unasserted assertions" is better than "unasserted facts".
<tl> how about "described triple"?
<niklasl> I can read "unasserted triple term" as a shorthand expression for "a triple term that is not also asserted" (i.e. is not also in the graph").
<niklasl> +1 to TallTed, annotation syntax seems essential to me.
<niklasl> I think Thomas means that if you roundtrip: `s: :p :o {| NAME e1 |} . << :e2 | :s :p :o >> .` we get `s: :p :o {| NAME :e1 |}, {| NAME :e2 |} .` I'm personally fine with that (since I expect it to work like that).