- From: Dan Brickley <Daniel.Brickley@bristol.ac.uk>
- Date: Tue, 30 Jan 2001 00:28:17 +0000 (GMT)
- To: www-annotation@w3.org
- cc: rss-dev@egroups.com
(Sent to W3C's annotations and copied to RSS-DEV; not sure where best to have this thread...) Just a quick note to share an in-progress experiment. At ILRT we've been looking at various forms of shared bookmark service, for ratings/recommendations etc. Eventually some of this code might get shrinkwrapped so others can run it; in the meantime here's a quick suggestion for an interoperability experiment. RSS (RDF Site Summary) is a data format for 'whats new' listings, news feeds and the like. The RSS 1.0 spec (http://purl.org/rss/1.0/) uses RDF to allow various other kinds of data (http://ilrt.org/discovery/2000/11/rss-query/ has an example) to be mixed into these "channels" of data. RSS is popular in the weblogs community, which is an application not a million miles away from annotations, shared bookmarks etc. So there's some shared ground between RSS and annotations that I'd like to explore as a practical 'semantic web' testbed. A couple of weeks ago (thanks to Libby Miller) I acquired a personal RSS feed corresponding to the shared bookmarks / recommendations I was adding into our experimental RDF metadata server. Rather than go into the details of that server (which is still being re-worked), it's enough to say that the data goes in via a WWW browser "bookmarklet" (Javascript fragment) and is exported using RSS 1.0. Bookmarklets are a neat hack; Netscape, IE (and I think Opera) browsers allow bookmarks to contain Javascript fragments that can solicit data from users when selected. See http://www.bookmarlets.com/ for details. Basically, when I browse an interesting site, I press a toolbar button in Netscape/IE, enter a comment and category into a pop-up window, and submit that to a database server for storage and subsequent access. The RSS channel at... http://ilrt.org/discovery/demos/RSSWeb/newitems-daniel.brickley@bristol.ac.uk/ ...is a general feed for recent bookmarks I've made. Other URLs give various subsets of the database (queries by date, topic etc.). There are some glitches (the sorting is currently wrong; as is the URI assigned to the channel) but you get the idea. While there are many ways one might represent this sort of data in XML/RDF, this is one simple way that seems to work. What I'm thinking is that regardless of the various bookmark/rating/annotation systems we're all using, a few of us could likely generate data in RSS format for some interop testing. This would work better for apps that use per-page annotations rather than annotatiosn attached to sections of a doc, though perhaps the idea extends to that too. RSS is interesting in that there are a growing number of tools (see http://purl.org/rss/) that consume RSS data. That's probably enough detail for now. There are a couple of links (with URLs to long to mail reliably) from my home page ([1]) to RSS viewers showing HTMLizations of the RSS feed mentioned above. What I'm hoping is to get a bit more data to play with from othesr on this list, so we could look at aggregating bookmark /recommendation feeds from multiple people with shared or complementary interests. Am I making any sense here? I could flesh out some details if anyone's interested in making such a testbed happen... all the best, Dan -- http://purl.org/net/danbri/ [1]
Received on Monday, 29 January 2001 19:30:26 UTC