Ian Horrocks spoke to the BCS on ontologies, the application of which would clearly see a true Semantic Web, but how can we apply these principles to the billions of existing Web pages?
Don't. Web pages are designed for people. For the Semantic Web we need to look at existing databases and the data in them.
To make this information useful semantically requires a sequence of events:
1. Do a model of what's in the database - which would give you an ontology you could work out on the back of an envelope. Write it in RDF Schema or OWL (the Web Ontology Language).
2. Find out who else has already got equivalent terms in an ontology. For those things use their terms instead.
3. Write down how your database connects to those things.
Using this information you can set up a Web server that runs resource description framework (RDF). A larger database could support queries.
To make all this really useful it's important that all important things - such as customers and products - have URIs (Uniform Resource Identifiers) - for example, http:// example.com/products.rdf#hairdryers - so invoices, shipping notes, product specifications and so on can refer to them.
These would all be virtual RDF files - the server would generate them on the fly and it would all be available on the Semantic Web. Then an individual could compare products directly by their specifications, weight and delivery charges, price and so on, in a way that HTML won't allow.
In your book you mention the aim of making the Web operate more like the human brain in making unusual and richer connections between data - doesn't the Web perform that function better in a way now because of the tangential returns you get from searches? Wouldn't the ontological approach make the Web less like the human brain?
Well, the Semantic Web is about data. The Web of human ideas is served by the hypertext Web but the Semantic Web helps with machine analysis. Take the current concerns over bird flu. Is it only around agricultural areas?
Suppose we have shared terms (URIs) for latitude and longitude and time. That would allow so many forms of public and private data to be correlated. We could also combine any say, medical data, with socio-economic data from the World Bank - land use and so on.
This could co-relate bird flu information and export it to a spreadsheet and lead to serious analysis. So, where HTML provides information in a way easy for humans to read, the semantic Web will enable much better analysis of it.