XDI and RDF Graph Model

Session: Wed Session 5 Space A

Conference: IIW 10 May 17-19, 2009 this is the complete Complete Set of Notes

Notes-taker(s): Markus Sabadello, Cameron Hunt

Notes:

I think nobody took notes, because pretty much everybody in the session was somehow involved in holding it :)

Well yes we went over some of the similarities and differences between XDI and RDF. Here is a quick list that is greatly simplified, but maybe it would be useful for the notes:

What the models share:
 * Semantic data, organized in triples, forming a graph
 * There are "resources" and "literals"
 * Various ways of querying and reasoning exist

Where they differ:
 * In XDI, there are semantics not only in the graph, but also in the identifiers (e.g. $v$1)
 * XDI has XDI messaging, RDF has SPARQL, Linked Data, FOAF+SSL
 * XDI has built in access control (link contracts)
 * RDF has blank nodes, XDI has inner graphs
 * RDF has ontologies, XDI has dictionaries
 * RDF can have multiple literals per subject/predicate, XDI only one
 * XDI can use persistent identifiers (i-numbers)

Regarding a bijective mapping between XDI and RDF, this is a topic that has haunted Drummond, Paul and me for years.. The short story is that a quick mapping is extremely easy (because in both models you just have triples), but the more "complete" you want the mapping to be, the harder it gets.

Such a mapping is actually implemented in several Higgins components. See here for a glimpse at this endeavor: http://wiki.eclipse.org/IdAS_XDI_Mapping  Markus

Cameron Hunt's notes:

Honestly, I started to take high level notes, but pretty soon I was jumping in (and up - using the white board). The discussion was pretty brief, and really just gave a high over view of XDI, then I responded with how those same things are being addressed by RDF-oriented solutions.

Specifically, while RDF alone certainly doesn't cover the broad capability set provided by XDI the Linked Data/SemWeb crowd is using RDF-oriented methodologies (SPARQL, Linked Data, FOAF+SSL, and even the proposals for RDF2) that claim to address those same sets of capabilities.

And while there are some pretty strong personalities that are difficult to engage, even those personalities are on record as claiming that the principles are more important than the implementation.

I myself (thanks to Randy) learned about the key weakness of FOAF+SSL (it doesn't separate the access token from the authorization token) - but I think there are some folks working on FOAF+SSL in particular and SemWeb/Linked Data in general that might be open to dialogue - I'm thinking in particular of Nathan (http://webr3.org/blog/) who is actively involved in those spaces.