Raw

Alex Barnett (formerly of Microsoft, now at BungeeLabs) joins a few dots : REST, ROA, MS's Astoria project, SOA & WOA. He quotes Jon Udell referring to the new book RESTful Web Services, by Leonard Richardson and Sam Ruby. The ROA is Resource-Oriented Architecture, WOA Web-Oriented Architecture (see Dion Hinchcliffe's post for more on this), both of which seem pretty much synonymous with REST, particularly when applied to data. Although neither Alex nor Dion mention this, REST applied to data is also what the Semantic Web is about. 

Alex wonders what these folks make of Project Astoria - I might as well give my €0.02 again. To all intents and purposes it's a Semantic Web system. Ok, they've got their own entity-relationship model rather than RDF, and their own approach to querying. While this could be written off as proprietary Microsoftisms, I believe that would be a mistake. Their internal model is pretty much isomorphic to RDF (as is also, for example, that of Metaweb's Freebase). The system is exposed over a RESTful HTTP interface - that makes it firmly part of the Web. In the Astoria demos they even provide RDF/XML representations. For better or worse, half the world is tied to MS systems, so I think it's entirely positive a SemWeb angle has emerged. ( It seems highly likely that longer-term the change in the environment that the web entails will force the fragmentation of monoliths like Microsoft, but that's another story...)

Yes, Astoria has its own URI query syntax, but conceptually it's very close to SPARQL. They've got bits like result grouping which aren't quite ready in SPARQL yet. In this case, rather than write it off, to me it would seem a good idea for SemWeb developers to borrow their URI query syntax and attach it to backend SPARQL+grouping extensions. Win-win.

Incidentally, I had a brief airport conversation with Jim Webber, who was skeptical about the Semantic Web on the basis that it was top-down. I was too knackered to reason, but did my best to give the arguments as found in the SW FAQ. But (rather ironically) I think Microsoft's Astoria adds weight to the argument that the Semantic Web is bottom-up. The common interface is HTTP, information/knowledge is represented in an entity-relationship form (which can be exposed as RDF/XML if required) - the approach would appear to pass the Test of Independent Invention... ( Duncan Cragg was at the airport too, also skeptical, but I plan to tackle his doubts head-on with code, once I've got my head around his 'Micro Web' stuff... btw, got slides Duncan?)