Re: Slides from Paul's talk

On Jun 9, 2010, at 4:50 PM, Melvin Carvalho wrote:

> <snip>

> Thanks.  I also read your blog post on this a while back, and Drummond Reed also blogged about personal data stores.

Yes, I'm working closely with Drummond (and others) on this. Using XDI of course :-)  
> 
> I do think it's the next big shift for the web ... and linked data with it's schemaless property I think is a good fit in this area.

No question. 

For example, the Higgins data model is based on the RDF model (with some generalizations and a few minor restrictions ). In fact we're having to work quite hard to map over some differences in the RDF-based Higgins graph model vs. the XDI graph model.

> 
> Kingsley Idehen has been talking about this for some time, and it makes a lot of sense. Also when SPARQL 1.1 comes out later this year, RDF data will be read/write as standard, rather than read only, which I think will allow a lot more applications to be built.

There are a number of things that we've had to invent in the past 6 years because they didn't at the time exist in what I might call the RDF stack. Here are a few:

globally linked graphs (now exists as Linked Data)
deep notions of "context" (we can mostly get there with named graphs, quad stores, etc, but still not easy)
fine grained access control (both policy description and enforcement) --this is an area that we've designed but (thankfully) haven't implemented yet. I've just recently read [1] with great interest. 
read/write access (sounds like that's coming)
pubsub data sharing --well, we're still working on this one (seems like a number of folks are working on this for RDF)
a loose correlation predicate (h:correlation) that is like a weak version of owl:sameAs
an ontology that describes people that builds on vCARD, FOAF, and any other useful vocabulary
how to describe change events --still working on this
how to describe provenance --still working on this
...plus some other stuff that RDF folks won't care about

Over the years as the RDF stack matures, we've been able to throw away the above inventions one by one. SPARQL 1.1 will allow us to throw out some more stuff. 

[1] http://www.google.com/url?sa=t&source=web&cd=1&ved=0CBYQFjAA&url=http%3A%2F%2Fdig.csail.mit.edu%2F2009%2FPapers%2FISWC%2Frdf-access-control%2Fpaper.pdf&ei=Bj8QTID1EYSdlgfq5-GPDQ&usg=AFQjCNFWn31tce6N4wtVBvPK1jcoGeSBJg
>