peter merholz briefly discusses one of the semantic web articles that i linked to yesterday and strikes a decidedly critical tone [ i’m quoting in full, since it’s not easy to permalink his writings until they’ve migrated to the archives ]:
“The
May issue of Scientific American features an article outlining the Semantic Web, Tim Berners-Lee’s latest drum to beat
(see the W3C papers on it here.)
The article explains:The Semantic Web will bring structure to the meaningful
content of Web pages, creating an environment where software
agents roaming from page to page can readily carry out sophisticated
tasks for users.This is one of those
Extremely Noble and massively complex endeavors wherein academics,
removed from the real world, attempt to solve a problem nobody
has. (I fear MIT’s Oxygen
project will suffer a similar fate.) The only reason the Semantic
Web gets any press is that Tim Berners-Lee, the "Father of
the Web," is working on it.History has shown us
that technology inventors often haven’t the faintest clue about
the device’s actual use. Did folks line up to hear what Philo
T Farnsworth had to say about television?Hypertext creators
tend to have Extremely Noble intents for their technology. Douglas
Engelbart was obsessed with "augmenting" intellect,
and the first development of the WWW was definitely along those
lines (for academics to compare notes). And the Semantic Web is
no different. From the SciAm article:If properly designed, the Semantic Web can assist the evolution
of human knowledge as a whole.Problem being, no one,
apart from some self-appointed Bringers Of Fire, wants their intellect
augmented, nor really cares about the "evolution of human
knowledge." The Web, this extremely exciting hypertext platform,
serves other human needs and desires–primarily to communicate,
also for sexual release (porn!), and for finding information of
personal relevance (what’s the weather where I’m traveling? how
can I do my job better? where’s my favorite band playing?).”
hmmm. i can’t tell from his rant if he’s against the ‘semantics’ part or the ‘assisting the evolution of human knowledge part’.
i have no gripe with the basic proposition that we need to formalize the semantics of both resources and services. this is a necessary requirement to support conversational technologies [ regardless if those conversations are taking place between humans or software ] and the whole web-based services scheme. the semantics are somewhere – either embedded in db schema, or an interface spec or in the natural language between the font tags in a web page. the only question is how one exposes the semantics to facilitate the location, discovery, description, and integration of resources and services. this, in turn, depends on the complexity of the system that you are trying to formalize. for “simple” software services, UDDI may suffice. for more complicated resources, you need more complicated ontological mechanisms. rdf might be right for some types of resources. maybe not.
“the semantic” web exists on a continuum of formalized semantics, with different methods being appropriate for various points along the continuum. at one end you have the status quo, with semantics embedded in “meta” tags, db schema and interface definition languages. at the other end you have rdf and stronger versions of knowledge representation. in between all this you get things like uddi.
the development of these multiple parallel versions of the semantic web are going to enable the very things that he thinks the market wants – like communication, sexual release and finding information of personal relevance.