Raw

Over on Richard MacManus's Read/WriteWeb, David Lenehan provides a good report on the London Mashup event, What's Next, Web 3.0?

[ PS. Another report, from Paul Miller of Talis.

Hmm, one part I'm not sure I agree with:

It's hard to add all the structure and metadata that many of today's Semantic Web applications require, the datasets are pretty small, and the use cases tend to be niche.

No harder (or more niche) surely than creating microformat data as consumed by the tools Sam Sethi demonstrated - because that's all perfectly good Semantic Web data. ]

Paul Walsh's material on Content Labels went down particularly well it seems, though it sounds like there were misperceptions in the audience about what the Semantic Web involves. David's example in the post made a handy jumping off point for a comment I posted, which I'm copying over here so I can find it again for probable recycling. I wish I could remember where I saw the quote (something like) which is the concise version of my comment: "What's new about the Semantic Web isn't the Semantic but the Web".

I get the impression there's maybe too much being read into "machine understandable" in the context of the Semantic Web. For the most part this "understanding" is simple logical reasoning down at the lower end of sophistication compared to what computers are already routinely doing. The difference is that it's done in a way that works on the web.

Ok, so you have a website that talks a lot about football, has no adult content, is child safe and is in French. Say you also have a potential reader that is 7 years of age, into sports and only speaks English. The machine can infer that the site's material is suitable for this reader in terms of child-safety, but probably won't be of interest because of the different language. It's easy to see how e.g. search results could be filtered in this way.

If the information is available, then another selection point there could be inferred : football is a sport, hence of interest to this reader. That classification may be provided by the source site, by the search service, by the reader or on a completely different site altogether.

The ability to express the statement "football is a sport" in a shared, machine-readable fashion (and the Content Label information), integrate this kind of information and allow things like filtering are facilities that the "Semantic" side of Semantic Web technologies provide. Being able to get hold of that information from diverse sources is part of the "Web".

Content Labels provide a great outreach point (as well as utility ;-), although they may inadvertently encourage another fairly common misperception, that RDF & Semantic Web tech are only about content metadata. So if I do end up recycling the above (heh, I guess this is now self-referential), I guess I should add the qualification that content metadata is just one use case, RDF can also describe real-world things and concepts directly - people, places, projects, events etc.

@en