Call for Participation: Hacking RDF with Perl Hackathon at Geilo, Norway, April 19th-21st.

[Gregory Williams]

On Mar 30, 2010, at 6:34 PM, Toby Inkster wrote:

> Here are some of my ideas:
> 
> 1. Greg has a subclass of RDF::Trine::Model that does forward-chaining
> RDFS reasoning. It would be nice if we could get this stable-ish.
> Possibly split it into a generic rules-based reasoning model, and a
> separate module that provides a set of RDFS rules. We could even build
> some other rules-providing modules, such as vocab-specific ones (e.g. a
> set of rules for FOAF would be able to generate foaf:mbox_sha1sum
> properties from foaf:mbox data).

Sounds like a great idea. That RDFS code has been bitrotting for a *long* time, and it would be nice to turn it into a stable framework for doing rule-based reasoning. The recent work on better support for quads might help here for trying to distinguish computed statements from the underlying graph data (this was a challenge I faced when I originally tried to write the RDFS code).

> 2. I'd like to build an ACL-protected subclass of RDF::Trine::Store.
> Data would be protected at a per-graph level (rather than a per-triple
> level). I've already got RDF::ACL working, but need help to get the
> get_statements, add_statement, etc methods of RDF::Trine::Store to
> respect an ACL.

Also a good idea. I had code trying to do that for a while in my SPARQL endpoint package, but trying to keep the endpoint and ACL stuff in one class was just unmaintainable. On that note, though, I'd like to package up RDF::Endpoint and get it released to CPAN (and update it to support the current state of SPARQL service descriptions).

> 3. Given #2, I'd like to build a SPARQL protocol endpoint that uses FOAF
> +SSL to establish the client's identity, and limit service to them
> depending on an ACL.

Nice. Let's sort out the intersection between the ACL Model stuff and my RDF::Endpoint code and get this working.


Another thing that come to my mind that might be nice to look into is adding native support for datatype canonicalization to both RDF importing and Query processing. Not quite all the way to D-Entailment, but would be nice to have.

.greg