Jon’s confessions on semantic cheating made me think that (as I explained previously) there are two ways to get away from semantic implicitness via augmentation: internal or external.

Jon tries the internal augmentation approach by adding more attributes to the elements and trying to find what is the best URI that describes me and yet keeps the XML tree clean enough to perform reasonably semantic xqueries on top of it using my name.

Interestingly enough, Jon is not alone in this: the RDF folks are using the exact same approach!! just they don’t care about tree readability and semantically meaningful xqueries since they have their own query language.

I’ve said this already, but the use of internal augmentation to exit the trap of semantic implictness strikes me as repeating the <FONT> mistake over again: before it was about escaping the styling implicitness, now it’s the semantic one.

But internal augmentation was proven evil for implicit style and I find it as evil for implicit semantics: adding a bunch of crappy and verbose RDF tags all over the place will not only make it far less readable and maintainable (just like the infamous <FONT> tag did for HTML), but will also reduce reuse.

So, if style was solved with external augmentation (stylesheets), why can’t we use external augmentation also for solving implicit semantics?

We are already doing this for Simile as we are using XSLT stylesheets to generate RDF out of XML corpuses, but XSLT it’s clearly not the right language for this and we should definitely make an effort to come up with something better.

Anyway, it’s nice to see other people discovering the limits of XML.

But it will be even nicer when people realize the need for external semantic augmentation instead of internal one.