earlier replies
Replying to @lomoramic @timFinin @ontotext
There is a tradeoff with the readability of the diagram. It's already a busy one
1
the diagram itself does not perturb me. the notion that it adequately conveys a real problem does. the same concern applies to the rdf-star report.
1
1
I think RDF* attempts to solve the problem that software engineers find LPGs more appealing because they can quickly qualify edges without giving much thought to domain modeling. RDF* allows one to more easily punt on domain modeling in RDF too.
1
1
RDF* is advertised as supporting LPGs but the CG report goes into a totally different direction: versioning, explainable AI, syntactic fidelity, triple types instead of actual occurrences. LPGs is the last thing the proposed semantics supports. So it will be ignored. One more...
2
1
RDF-star already covers very well what developers are used to do with edge properties in LPG. I don't feel like judging whether that's good or bad thing to do. The CG dives deep into many semantic aspects - something people in the RDF camp us to do. I don't see this harmful
1
"harmful" is not the question. try "useful". - does anything more than "what developers are used to [doing] with edge properties" - that is the "deep ... semantic aspect", matter? - are the real things one does with rdf supported?
1
2
Yes, @Synaptica uses @OntotextGraphDB support for #RDF-star for access control management in their Graphite knowledge management suite. The benefits in terms of simplicity of modeling and smaller storage footprint are clear. ontotext.com/blog/rdf-star-i…

RDF-star Implementation in GraphDB and How Synaptica Used It Within Graphite for Access Control

Synaptica's Gene Loh and Ontotext's Vassil Momtchev discuss RDF-star, its value and how it was implemented in Ontotext’s GraphDB and Synaptica’s Graphite.

ontotext.com
2
2
Again unasserted assertions. Really, before the RDF-star CG I had no idea that unasserted assertions are such a thing among practitioners. Why did the proposals for an RDF literal datatype didn't get more support? Much more straightforward encoding of unasserted assertions IMHO
1
The problem with modelling quoted triples as literals (with a dedicated datatype) would have been blank nodes. You would not have been able to model something like : <#alice> foaf:knows _:bob {| ex:since 2019 [}. _:bob foaf:name "Bob".
1
Or would you? Isn't the example rather a case for Quoted Graphs? "Blank nodes belong to _one_ surface, end of story". Then i.e. skolemize into RDF-typed literal, end of story. No trouble with semantics. Your example above has wiggle room for strangely differing interpretations.
2
Replying to @tomlurge @kiryakov_ak @lomoramic @thejustman @timFinin @ontotext @Synaptica @OntotextGraphDB
> Your example above has wiggle room for strangely differing interpretations. AFAICT, the RDF-star report unambiguously specifies what the standard interpretation is. w3.org/2021/12/rdf-star.html If it does not, it needs to be fixed, which will be the job of the working group.

Sep 2, 2022 · 5:31 AM UTC

1
Replying to @pchampin @kiryakov_ak @lomoramic @thejustman @timFinin @ontotext @Synaptica @OntotextGraphDB
What I mean is that it's strange to say: "Alice knows someone since 2019. Someone is named Bob." It would be more natural to say: "Alice knows someone named Bob since 2019" but that would require a quoted graph. Annotating the outer triple in a n-ary relation however works better