Raw

Some great slides: Semantic Web - A Quick But Detailed Introduction



Great, not only because he give my stuff a hat-tip (heh) but also because he ties together some outward-facing developments: microformats, embedded RDF (eRDF and RDFa), GRDDL, SPARQL.



Classic line:

Americans learn quite quickly never to eat anything that British people offer them. Marmite especially.

He's also been doing a fair bit of work with connecting between OPML and semweb tech, one of those things that IMHO should be done but is in the category of 'rather you than me'.



He's also set up services for Tidy and XSLT, allowing URL-chaining to get from microformats to RDF (approximately the GRDDL idea). Very cool, and gives me something to talk about in the context of simple service descriptions, but I need to catch up on the Linking Open Data list before going any further with that.

I just listened to his podcast - very listenable and gets to the nitty-gritty, asking (after the 7 minutes of intro...it is his first podcast): " How is that useful for me today?". Nice insights on del.icio.us and folksonomies, countering the metacrap notions, moving via machine tags to the RDF model.

I was just about to give Big Respect to Tom for the music he played out on - except it wasn't Tom, it was just a (magic) coincidence:

~

One remark from Tom on the slides I disagree with, looking ahead he says : " Hopefully, less 900 page specifications and less waffly academics."



But I like 900 page specifications and waffly academics! The waffly academics cover the theoretical work so I don't have to, and the 900 page specifications mean I can get the answers I need.

Don't get me wrong, if I wanted to learn how to create a feed in an afternoon, all things being equal I'd be better of reading something like the RSS 2.0 spec than the Atom spec. That's because it's not really a specification, it's a primer . But it is entirely possible to make a primer to support a specification, here's one for Atom. The difference is, if I get stuck on a detail ( can an item/entry include more that one enclosure?) I can look it up. Without a real spec, all I can do is guess and hope everyone else on the web made the same guess... 

The state of the art for specification really has advanced in the past few years. Software is stuff that interfaces between people and machines, and for standards to work well their definition should reflect this. So for example the material for GRDDL covers the scale from human-friendly to machine-readable. The package includes:

• Primer (overview, what it does - human)
• Use Cases (why you might want to do it - human)
• Specification (how it should be done - human/machine)
• Test Suite (check if you're doing it right - machine)

The tests are currently for GRDDL tool implementors not publishers, but all being well there'll be a validator very soon. (There's a fairly tight feedback loop from implementors like those of Jena and Raptor back to the Working Group).

Still minor issues to clear, but it's heading rapidly towards Last Call...

@en