RDF*/SPARQL* support?

[sa-bpelakh]

There is currently a W3C working group considering the adoption of Olaf Hartig's work on RDF*/SPARQL* as a W3C standard (http://olafhartig.de/files/Hartig_AMW2017_RDFStar.pdf). Several DB vendors already support this syntax (Anzo, Stardog, Blazegraph), and others are working on adding it as a feature. It provides a great bridge between RDF and Property graphs.

Could this be possibly added to rdflib?

[gromgull]

Since RDFLib has essentially 0 active developers and we don't even manage to make a release - I don't see that as very likely :D Unless someone contributes a very good PR ! - Gunnar

…

On Thu, 6 Feb 2020 at 20:02, Boris Pelakh ***@***.***> wrote: There is currently a W3C working group considering the adoption of Olaf Hartig's work on RDF*/SPARQL* as a W3C standard ( http://olafhartig.de/files/Hartig_AMW2017_RDFStar.pdf). Several DB vendors already support this syntax (Anzo, Stardog, Blazegraph), and others are working on adding it as a feature. It provides a great bridge between RDF and Property graphs. Could this be possibly added to rdflib? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#955?email_source=notifications&email_token=AABLO3F3OTAZ3534CIEUQFDRBRNEVA5CNFSM4KRCIRJKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4ILTLQTQ>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABLO3CSUSLR4UK3V5Y5CMTRBRNEVANCNFSM4KRCIRJA> .

-- http://gromgull.net

[JervenBolleman]

@gromgull I am working on it! Hope to publish a branch next week.

[sa-bpelakh]

@JervenBolleman that is fantastic, please let me know if you need a hand with testing.

[JervenBolleman]

The approach I am taking is by going the RDF* PG mode where <ex:p ex:y ex:z> asserts the triple and itself is a special BNode (basically a bnode with a derivable _:id) Which maps to reified statements. Has it's downsides but I think making RDF* a syntactic sugar is the way to go. At least with my limited skills.

[JervenBolleman]

Hi, just wanted to note that my slow progress is available here. I am however, having difficulty with the parser code. So I was very over optimistic on getting this work faster.

[JervenBolleman]

I am now into adding the new EmpTP nodes into the algebra. Recursive <<?s ?o <<?s2 ?o2 ?p2>>>> does not work, as I don't know how to define that in the parser logic.

[JervenBolleman]

Well "it works" now. PG mode, SPARQL only by translating the sparql* into a reification quad.
Needs serious work to polish it all up and make it more Pythonic.

[nicholascar]

Great work here @JervenBolleman! We (my company) will likely be looking at RDF* soon for W3C graph work (GQL).

[JervenBolleman]

@nicholascar thank you. Still a lot of work to be done.

• Move EmbTP to reification quad translation into evaluation from algebra (Via new triple object)
• Implement SPARQL* update
• Make SPARQL* recursive in the parser
• Add many more test cases!
• BIND and VALUES need to work as well.
• Figure out why URIRef inside the <<>> are not transformed into URIRef objects

[JervenBolleman]

Just want to share my progress. The parsing is done better now, and instead of hacking in support to the algebra it is now pushed down into the evaluation. However, I have not yet managed to get the evaluate to work nicely. It is natural to put it in the evalBGP part, but resolving an "EmbeddedTriple" to a series of Identifiers is quite painful there. I need to think a bit more.