Re: Announce: Browse RDF like the web

Very nice!  Some people are skeptical that generic SemWeb browsers
will be very useful, but I think its definitely worth a try.  (Now I
just need to update my Java installation....)

I did a little work on something similar (SemWalker [2]), which left
me with some ideas you may want to try out:

  +  Use the thing's home information (the content available at the URI
     part of the URI-Ref) by default, so users don't always have to
     name a source.  It's kind of like trying ibm.com as a web address
     before Googling for it.

  +  Display backlinks.  The browser from Mike Dean et al (in SONAT)
     does them in a different color; I like the idea of displaying
     them on the left, so the page ends up looking like a narrow view
     of a circles-and-arrows diagram.  I did a mockup last december [1]
     which may make this clearer.  (I suggested this to Mike et al a
     few weeks ago; we'll see whether they pick it up.)

  +  Have a mode which generates an HTML page with several of these
     object displays, each one tagged with an HTML id matching the
     fragment portion of the URI-Ref.  Such a page could be published
     as HTML at the home address of the group of things, so people
     asking (via content-negotiation) for HTML (instead of RDF) at the
     URI will get something pretty.  Write the links out of that page
     to go through a public BrownSauce server.  (All such servers
     should respect a cookie indicating the user would like to be
     redirected to a different server-side semantic web browser.)

  +  Oh yeah -- provide a public server, at least for experimentation!
     There's no need for me install Java after all.

  +  Keep aggregation fairly separate; an aggregating source is just
     another source, and should be usable outside of brownsauce.  I
     imagine something like 
            [ ] just use home information
            [ ] use information from source: ________________________
            [ ] start me an aggregator.

     If they're using an aggregator, then give them some UI to control
     it.  It'd be nice if the aggregrator's essential state could be
     kept in its URI, so that it appears to live forever even if it
     gets reaped after 10 minutes of idle time.

Wishing he had more time to work on this himself,

  -- sandro

[1] http://www.w3.org/2001/12/server-side-RDF-browser/mockup1.html
[2] http://www.w3.org/2002/05/semwalker/
[3] http://www.w3.org/2002/08/LX/

Received on Thursday, 7 November 2002 08:10:23 UTC