Pragmatism and the semantic web

Paul Miller at Talis mentioned a Tim Berners-Lee video in which he tries to explain, in plain English, the semantic web to ZDNet's Executive Editor David Berlind.
He chose the term 'data bus' to describe the end result. Computers can dip into the data bus and drag out whatever information they need, providing it's in there, and manipulate it how they see fit.
The data gets on the bus by being encoded using the RDF standard data format. And it can be sucked out using a query language called SPARQL, among other things.
(These are all my words, by the way, not Sir Tim's. But I hope I'm reflecting a new mood of pragmatism that Miller was relieved to note when Berners-Lee was talking at last month's www2007 conference in Banff. )
Changing existing data formats would prove a nightmare within and between organisations. They would have to trickle the changes to the furthest reaches of their organisations and their partners. A better approach for existing data is to 'wrap' it inside RDF. And this is how the great migration to the semantic web is going to happen.
It will start with data which benefits its owners from being open. It will be sucked up by computers belonging to people who believe they can deliver value by linking disparate information together.
Think of mashups. They link, say, mapping information with building information and, maybe, deeper level stuff about who's in the building and what they are doing there. Okay, I made that last bit up, but it's feasible. The main issue is that the mashup writer has to explicitly link up to the APIs for what they want to surface. These might be proprietary and, in any event, once a mashup is done, it does only that job.
Apply the same thinking to the semantic web: the data is in standard formats (think XML) and it can be grabbed and manipulated by software. Not only that, but relationship information and alternative terms can be embedded and defined.
David Tebbutt Likes IWR, for example. A subject, predicate and object. Such encoding, derived from one of my reading lists or my vote in StumbleUpon could be extracted and aggregated with others to draw new conclusions. Expect to see provenance and/or trust attributes as well.
The semantic web project is under way. Companies like Talis are already working hard to make sure that their own products and services exploit these standards to go way beyond the narrow world of libraries.
The W3C stuff makes for dry reading, but we all owe it to ourselves to become familiar with the opportunities presented by the sematic web.