- From: Richard Newman <r.newman@reading.ac.uk>
- Date: Sun, 27 Mar 2005 23:42:54 +0100
- To: Stan James <sjames@uni-osnabrueck.de>
- Cc: semantic-web@w3.org
> Sorry I didn't mention the application: My bookmark processor acts > recursively. So if I've bookmarked your bookmark page, then I can get > your bookmarks too. But it needs to know which items it should/could > recurse on. In which case, you're essentially wanting to infer some behaviour based on the type of the bookmark. There are only three ways of doing this, and two of them are the same. Those two are: 1. Infer that it's bookmarkable content by retrieving the URL and parsing the returned representation. 2. Do the same, but pre-encode the knowledge by using a different property. The other one is: 3. Explicitly encode the behaviour through a tag like "importbookmark". This is fine to me; as long as you don't treat the annotation as being intrinsic to the data. Note that 1 is actually better than 2 --- I may have a bookmark service that returns HTML or RDF depending on the Accepts: header of the client. You might not know that until you try to fetch it; the key thing is that your software works with representations, not resources. It seems to me that the recursive action should be part of the program (based on an analysis of returned representations), not the data, but I'm not in full possession of the facts. :) -R
Received on Sunday, 27 March 2005 22:43:32 UTC