Lou Rosenfeld has his take on folksonomies, referring to them as metadata ecologies. It’s a good log on public-driven metadata and the issues surrounding it. *sigh* I was hoping for a nice solution from Lou…but maybe I haven’t defined the problem well enough…
Anyway, the point I think he makes about Flickr’s metadata scheme topping out and becoming less and less useful for the public to retrieve content is absolutely correct. For example, on Flickr’s largest tag, 2004, there are currently 119,103 photos. That’s a lot to dig through. I also think that people will catch on to the unique tagging idea, and pretty soon there will be license-plate tags cluttering up everything.
His article, like so many others, gave me more to consider as I wrestle with my own categorization problem. I put categories on everything, not just on-line data. While I have categories for my Furl archives, Bloglines subscriptions and Flickr photos, I also have schemes for the photos on my hard drive, e-mail archives, etc. etc. etc. I need something to tie all of this data together. A desktop search application isn’t the answer — most of my content lives on-line. Internet search isn’t the answer either because I have content on my home computer, work computer, PDA. All of it needs to be organized and retrievable.
So, what to do, what to do? I need metadata that’s convenient to use, easy to update and add to, that cascades changes throughout my data, and is integrated into the applications that use it. RSS may be a way to deliver that. Centrally locate an XML version of the metadata and have my applications poll it for updates? I know I’m ignoring the obvious problem of building a sustainable set of metadata, but I think it’s a sound idea. The technology exists. And my metadata only needs to make sense to me, really.
Hmm…and as I continue to ponder Lou’s last question about how to marry the best of public-driven metadata with professionally organized and maintained metadata…I wonder, what about a suggestion engine combined with search engine. I might spew out something like: “People who searched on that term also searched on these variations…”
Okay, there’s a few ideas muddled and mashed together. Sorry about that, it’s where my head is today. If you’re a genius and want to build this for me and talk Bloglines, etc. into integrating it into their products, please let me know. Everyone needs a hobby.
You and Lou (among others) must be on the wavelength of sorts. The Jan 9 Bloug post deals with cross-silo content inventory and classification. I know it’s not exactly what you refer to but it is part of the problem you refer to: http://louisrosenfeld.com/home/bloug_archive/000331.html