Nice article from Bill de hÓra about why he's using RDF and the advantages of doing so.
This reminds me of the bit of my XMLUK talk that I unfortunately didn't get round to doing -- ran out of time due to too much enthusing about XForms.
Basically I was going to take a stab at a sales pitch for RDF but one that tried to undercut most of the hype associated with the term "semantic web". A pitch that a programmer could more easily buy into. Comments welcome.
It went something like this:
If you adopt a common, underlying model for your data (the graph) then you don't have to worry about data storage and persistence mechanisms: you can leave the rocket scientists to design and optimise that for you. Just like you do now with an RDBMS, with the advantage that you don't need a DBA to design your data structures for you.
If you've got a common, underlying model for your data then you can access it uniformly. No need for different APIs for different types of data, and again you can employ a rocket scientist to design a query language for you, and it'll work against all your data.
If your underlying model has built-in support for merging properties, captured from multiple applications, that refer to the same object, then there's one less integration head-ache and one less source of application model changes E.g. the need to extend object models to add mapping into and out of different representations, annotate objects with extra data (or metadata), etc.
If your model had a normative mapping to XML so that you still gain the interoperability benefits of an open, Unicode compliant format, and your toolkit will do it for you (no more worrying about XML to object mappings!)
If you can write a declarative schema to help merge and relate your data to and from other sources then that's less code you need to write, and opens the possibility for many more interesting applications.
...and if it just so happens that whilst you've been playing with this other rocket scientists have slapped rule engines and reasoning frameworks on top of your model, because just like SQL it has formal underpinnings, then you can start to explore all sorts of cool features. Or not. All of the preceeding advantages still hold even if you don't use this aspect.
Wouldn't that be a nice environment to work in?
Any RDF hackers out there want to add anything?
Posted by ldodds at April 28, 2004 03:28 PM | Feedback? | | TrackBackI tried this advocacy approach last year - avoiding all the semantic web fluff and focussing on RDF as a data model (http://www.semanticplanet.com/2003/04/rdfForModularExtensibleMarkup.html)
Posted by: Ian Davis on April 28, 2004 04:13 PM"Any RDF hackers out there want to add anything?"
The two analogies that seem to working for me lately. First, for WS-types; compare RDF to having a standard messaging envelope. Envelopes get you extensibility, uniformity, tracking, and so on. RDF gives you similar kinds of leverage for data. Second; everyone does property-value pairs, RDF just asks to name the thing the pair applies to and where possible use global keys. This seems to make sense to programmers.
I'm no RDF hacker, but your pitch, Ian's article, Bill de hÓra and Mark Baker's favourable comments convinced me to take a very close look and I liked what I saw, so I'm on the road to becoming one!
Would you care to elaborate on an RDF toolset to match the pitch?
Posted by: robert on May 4, 2004 12:03 PMI'm a bit late commenting on this one, as I've been busy getting a new product build ready, but now that that is out of the way I think its worth chiming in with this:
It _is_ (not just would be) a nice environment to work in, and not just for semi-theoretical "out there" projects - my company has used RDF in exactly this capacity for the last two years to power our data integration product at every level, from configuration files through schemata for processed data right on up to management information.
Too often I've heard RDF pundits say (or being said to have said) that "the tools will save us." Too often I've heard RDF haters use the phrase against RDF. Its time both parties recognize that the tools are here, they are real, and they work very well indeed.
Heck, I'd go so far as to say the tools are generally _mature_, assuming that our two years of use in a practical real-world business application counts towards maturity.
Posted by: Jeremy Gray on May 6, 2004 09:05 PM