|
Excellent reporting; thanks for keeping us well informed.
I had written a few thoughts about why RDF hasn't (yet) achieved its potential - I was going to put these in my blog, but I think this may have a better audience here:
* Commercial Incentives
Organisations adopting early HTML had obvious commercial incentives – an extremely low cost, low risk route to increased marketing, customer interaction, and additional sales channels.
XML presented more cost-saving opportunities – share data more easily with other organisations, migrate data across applications and platforms, and create content once for multiple delivery channels – "future-proof your data!"
RDF though, is harder to sell: "So – let me get this straight... If we invest thirty man days into creating and publishing RDF data, our customers may be able to find us on a search engine that may exist in the future which may be better than Google? Sign me up!"
Of course, RDF is much more than search-engine metadata – it's the data equivalent of Kazaa, BitTorrent, or any other de-centralised, peer-to-peer architecture which brings power to the people.
* Killer Application
Which brings us to the Killer Application, or lack there-of. For HTML, there was e-commerce. For XML, there was XSLT. RDF needs an equivalent – something that uses RDF in such a way that you just have to use it – without it, you're losing out to your competitors.
Perhaps it will be the RDF enabled search engine. Perhaps a popular, RDF-aware P2P application. Or perhaps Microsoft will integrate RDF-aware tools into Internet Explorer or Office applications.
Whatever it is – it doesn't exist - yet. That isn't to say that RDF has no current uses – FOAF, RSS, etc. – it just doesn’t possess a must have application.
* Understanding
If there's one thing worse than a Reese Witherspoon movie, it's hearing a respected technical team member explain to their peers that "choosing between RDF and XML" is arbitrary, as if the two are interchangeable syntaxes.
I've witnessed the "We’ll use XML now and convert to RDF later if necessary" conversation a few times – the model and implications of the two either side-stepped or, more usually, not understood.
It's easy to appreciate why this happens – metadata models often have separate 'XML' and 'RDF' bindings, reinforcing the stereotype of RDF as an XML alternative. Similarly, most developers will be aware of XML Schema and RDF Schema, the names of the two specifications sounding so similar that surely they are alternatives for the same domain?
If it's difficult for technical users to grasp the RDF model, then highlighting the benefits of RDF to managers and senior decision makers is verging on impossible.
* Software for Imbeciles
Let's be honest about RDF – the model is confusing, useful data is difficult to create, and the learning curve is steep. In some ways, it's similar to assembly language – a simple model and syntax that becomes more complex as you build useful, larger applications with it.
High-level, user-friendly editors are therefore essential for the widespread adoption, and ultimate success of RDF. However, the majority of RDF editors available today continue to confuse the user with predicates, reification, ontology editors, 3-dimensional webs of orbiting triplets, and other low-level data and terms.
Most users would just like to record some basic information about their web page, without any mention of statements or unique resource identifiers that may be created or implied from their original input.
* Press Coverage and The Buzz
XML was - and continues to be - warmly received by the press. Before XML was widely understood, magazines and newsletters would wax lyrical on the new XML technology, which would solve every I.T. problem from the last 20 years. Even now, additional minor XML features from major vendors (e.g. Microsoft, Macromedia, etc.) are widely publicised. Conversely, RDF functionality (e.g. Adobe Acrobat's support for Dublin Core based RDF) rarely excites journalists.
Without a buzz of excitement, the press will ignore the technology. Without the press, commercial decision makers will remain unaware of the technology, and therefore uninterested in its implementation – locking RDF into a minority 'geeks and nerds only' club.
* Common Classification System
Due to a lack of a central, widely-adopted common taxonomy (classification system, restricted vocabulary, whatever you want to call it), even current RDF data isn't living up to it's abilities. We need some common unique identifiers for common terms (words, topics, people, places), so that the real 'web' of information can begin to interlock together.
Creating and maintaing these identifiers, and keeping them non-proprietary, is an essential yet mammoth task.
|
Previous Message
