http://norman.walsh.name/2005/02/20/tagback 8 26 2005-02-20T12:47:20-05:00 $Date: 2005-09-11 10:27:02 -0400 (Sun, 11 Sep 2005) $ Norman Walsh 2005 Norman Walsh I've added provisions for folksonomy tags and plan to adopt them as a replacement for threads. The web has been buzzing with discussion about ontologies recently, and especially about folksonomies. Since the very beginning of this blog, I've been keeping track of metadata about each entry, including how each post fits into my own personal taxonomy. A personal taxonomy (akaAlso Known As folksonomy) is all well and good. You can use it, for example, to see the topics index on this site. But real value is only going to come from it if other people start to use it. Realistically, the chances of other folks using my ontology for their posts is remote at best. But with Technorati now building tag-based indexes of blogs, Flickr photographs, and del.icio.us and furl bookmarks, the possibility of hooking into a shared ontology seems much greater. To that end, I've reworked the taxonomy of these pages to support tags, per the Technorati specification. I haven't removed my own taxonomy, just provided an additional mechanism for public tagging. Following Shelley Powers ’ lead, I'm also using adding unique “tagback” tags for specific threads of discussion. This will (very quickly, I think) replace the poorly maintained threading system that I have been using. The biggest problem with the threading system is that only I can update the threads. With the folksonomy tags, anyone can add their own essay to the thread or use del.icio.us or furl to add someone else's. Only time will tell how well this system stands up to use and abuse.

As regular readers may have already guessed, I've abandoned my Palm device in favor of my new Sidekick. The hardest part of this transition (beyond some of the irksome shortcomings of the device) is rebuilding all of the server-side tools that I used to use to manage my personal information. One of the aspects I'm struggling with is the whole RDF thing. I'm strongly tempted to just use Plain Old XML. Frankly, it'd be faster and somewhat easier to do it that way. At least at first. It's much less clear that POXPlain Old XML would be the best long term answer for this particular application. That point was driven home dramatically when I set out to update this site to use tags. I accomplished the entire task by adding a few new bits of metadata in the taxonomy and the topics files, by adding a single rule to the set of inferences used by cwm to build the metadata for this site, and by updating one template in the stylesheet that builds the HTML. It's hard to imagine that it would have been just that easy if I'd only used POXPlain Old XML and custom scripts to build this site. RDF's ability to aggregate disparate sources of information and to be extended by inference shows promise once again. (In case you go looking for that rule, you won't find it anymore. After I had it all working, I decided to do it a little bit differently, changing the way topics and tags are processed so that they're generated dynamically by the server instead of being baked into the HTML. That required moving things around a little bit and made the inference unnecessary. But the RDF is still used to build the runtime taxonomy file and it's still just a simple matter of adding a few new RDF statements to build new tags and new taxonomies.)