Saturday, October 23, 2004

Who owns subscription data?. As Bloglines continues to grow, its subscription data becomes more interesting. Yesterday, while demonstrating the application to some folks, I showed how you can navigate from a feed to the list of subscribers to that feed. Only those subscribers who agree to make their data public are shown. So in my case, for example, there are only 755 public subscribers out of 2337 total subscribers. But that's more than enough transparency to do some useful data mining. ... [Jon's Radio] 12:43:35 PM      Google It!.

The Handle System. A comment in my discussion area yesterday noted that, for the Koper link, "they actually use the Handle system and state explicitly to point instead to http://hdl.handle.net/1820/238" (or hdl:1820/238, though most browsers don't support that notation). The idea of the handle system is that, instead of pointing directly to a link, you point to a handle proxy server, which will redirect you to the link. This is exactly how Persistent Url (or Purl) works, except without the special handle syntax. The benefit is that if the URL for the resource changes, links using Handle still work (assuming that someone remembers to update the Handle proxy server). On the other hand, as the Handle documentation states, the system is perfect for managing Intellectual property - that is, it is perfect for enforcing access restrictions at the network level, rather than at the server or resource level. Maybe so - but this was the point of my discussion with Dan Rehak in Utah over CORDRA, which also uses the Handle system. Applying digital rights at the network level, in addition to imposing a substantial overhead on everybody, runs the risk of fracturing the internet into a series of private networks. Will CORDRA or Handle do that? I don't know. I have no doubt that some people would like it to, though, and so urge caution regarding the implementation of Handle. The Handle system is used now by Digital Object Identifier (DOI), Defense Virtual Library, and DSpace, among others. By Various Authors, October 22, 2004 [Refer][Research][Reflect] [OLDaily] 12:41:37 PM      Google It!.

Metadata for the Masses. Nice brief discussion of the idea of ethnoclassification - classifications systems evolving through undirected use by a population, rather than stipulated from above by a standards body. The path analogy is especially apt. By Peter Merholz, Adaptive Path, October 19, 2004 [Refer][Research][Reflect] [OLDaily] 12:40:21 PM      Google It!.

Describe This. According to the announcement, Describe This is "a service designed for the automatic extraction of metadata from online resources. The site offers an easy to use interface where you can indicate the resource to analyze and how to download the results as XML, XHTML or RDF files." I tested it on my own website and on David Merrill's PDF (see below) and it worked for both, though certain desirable metadata (such as DC:creator) were missing. It can automatically analyze and generate metadata registers for the following formats: HTML and XHTML, Dublin Core/RDF, Dublin Core/XML, Dublin Core/HTML (META tags), GIF, JPG (EXIF) and other image formats, RSS, bibTex and some proprietary Formats XML, for example Amazon XML Web Services. Thanks, Toni, for the link. By Various Authors, Sand's Dublin Core Services, October, 2004 [Refer][Research][Reflect] [OLDaily] 12:38:52 PM      Google It!.