Saturday, March 10, 2007

Freebasing the Web

TechCrunch covers a new resource called Freebase that hopes to use the folksonomy as the basis of semantic web applications rather than the intricately crafted taxonomies favored by the World Wide Web consortium (W3C).

While the W3C's proposal makes a lot of sense and is how any rational person would approach the project, the failure of such simple initiatives as Dublin Core metadata makes me think that tagging and folksonomy is the best we're going to have to work with and if artificial intelligence clients can build a "crosswalk" to relate many of these folksonomies topics, then maybe we'll end up with something useful in the end.


dr said...

" how any rational person would approach the project..."

The problem with W3C's taxonomy, of course, is that this assertion is wildly false. Rational persons approach taxonomical sorting in lots of different ways depending on their perspective. W3C's taxonomy fails because it privileges one rational perspective as the rationality. That can only work in an environment where gatekeepers have power to control behavior. Given power to opt out, users do.

Safety Neal said...

I don't agree entirely. My understanding is that the W3C's approach is essentially modular. If someone has created a bit of taxonomy that is useful to you, you plug it in. If it doesn't work for you, you can plug in a different taxonomy or create your own.

The design doesn't require one single overarching taxonomy but allows for a different number of taxonomic components to be used together.

To my mind the system fails not b/c of the need for gatekeepers but b/c of the programmer bottleneck.

The W3C's approach needs lots and lots of programmers making these complex taxonomic components (to spec) with no tangible reward in sight for all of their effort.