Month: June 2005

As you may, or may not, have noticed, I’ve been integrating various pieces of metadata into the site, primarily into each individual post. Eventually I’ll remove the ‘meta’ option for each page, and just provide a machine consumable RDF/XML option — the humanly readable components will show directly.

Right now I’m creating a plugin in that pulls the links within a post and adds these into the metadata as references associated with the post. I’ve also added a form that allows me to add references that aren’t linked in the post (and which are printed out at the end of the post).

I’m playing around with tags and other odds and ends, but I wish I could find the right combination of data and implementation to really effectively demonstrate how all this holds together.

I’ve been incorporating the semantic data my application gathers into weblog posts. You can see it in operation over at Burningbird, in the individual posts (see references example below and the photo example).

During this, I ran into a wall on the topic of tags. I wanted to record tag-like information as RDF statements, but then I realized that I don’t necessarily know what tags are.

According to de.licio.us and furl, tags are ways of publishing bookmarks to a broader audience, in addition to categorizing your links. Since the sites are for bookmarking, adding links to your own work is frowned on.

In Flickr, tags are ways of categorizing your work, pure and simple. You may specifically use certain tags to participate in a community, but the majority of use is to classify one’s own work.

In Technorati, though, which is the one I’m most closely examining, a tag is a way of classifying your work for some purpose. According to the Technorati Tag instructions, you can link it to a Technorati page, but you can also link it to a Wikipedia or other page. However, according to the Technorati Wiki a tag is meant to reference a page that will aggregate the results (this is a wiki, note that text just quoted is subject to change). And therein lies the confusion about ‘purpose’.

If tags used in the sense that Technorati uses them are meant to help aggregate content actively, then yes, there needs to be specific pages and/or sites for the target URI–ones that actively gather and than republish incoming links.

However, if tags are meant to be more passively consumed, with bots going out and gathering the information, than as long as an agreed on format is used, any page can be linked (well, as long as you don’t link the same URL twice in the same document — Technorati sees this as spam).

I can’t map ‘tag’ into the semantic webspace using RDF if I can’t find a common meaning between all these distinct uses of the concept of ‘tag’. I spent time last night with this, and again this morning, but nothing fits.

I noticed that Norm Walsh used the relMeta wiki page as a namespace for a tiny self-contained schema reflecting ‘tag’ he uses in his taxonomy. That’s an option, I guess. But then, does that mean the Technorati namespaced schema doesn’t apply to de.licio.us, furl, and Flickr?