- From: Danny Ayers <danny666@virgilio.it>
- Date: Wed, 17 Jul 2002 21:15:03 +0200
- To: "Damian Steer" <dsteer@hp.com>, <www-rdf-interest@w3.org>
I've been working on this problem for a similar purpose, though haven't come up with any magic. What looks most promising to me so far is to take advantage of the topology of the graph (something you shouldn't have too much trouble with after RDFAuthor ;-) The wrapping ('folding' is the term I've been using) might start at any particular point in the graph, and examine the neighbouring nodes. If it finds something that can act as a terminus (a non-bNode resource) then it stops there and carry out the same test in different directions, until a little cluster that can stand alone is formed, which may be given an ID. This can be applied recursively, applying the same approach to the terminals of the cluster to produce a larger grouping (of course you don't need to give it an ID until you've got everything you want). kindaLike striping in circles. The problem of how you know you've got everything you want in the box is the kind of thing a query language might be good at, and the ID issuing could almost certainly be simplified using something like Seth's quads, but I reckon both of these would depend a lot on other requirements/facilities of the app (simply adding another property to each of the members of the cluster would be an alternative). Cheers, Danny. --- Danny Ayers <stuff> http://www.isacat.net </stuff> Idea maps for the Semantic Web http://www.isacat.net/ideagraph >-----Original Message----- >From: www-rdf-interest-request@w3.org >[mailto:www-rdf-interest-request@w3.org]On Behalf Of Damian Steer >Sent: 17 July 2002 13:47 >To: www-rdf-interest@w3.org >Subject: Objects as RDF subgraphs > > > >-----BEGIN PGP SIGNED MESSAGE----- >Hash: SHA1 > >Hi, > >I'm about to start work on an(other) rdf browser. The idea is to >present rdf data in a 'coarse grained' way - eg vcard data as a box >containing name, tel no, photo etc. > >The tricky bit will be the coarse graining. I'm considering >auto-generating classes which will pick out sub-graphs of an >underlying rdf model. > >The question is - how? It seems like the best thing to use would be an >rdf query to a) express what subgraphs the class 'wraps' and maybe b) >represent the fields of the class. > >eg for rdfDB [1] style queries the 'WHERE' clause corresponds to a) >and the SELECT clause (maybe) gives b). > >I'm a lazy guy, so is anyone else doing something similar, or know of >similar projects? I've been looking at neoom [2] (thanks keo) and MDL >[3] (thanks danbri). > >The problem of wrapping seems to crop up quite alot in rdf apps so any >work would help poor hackers. It also seems to relate to data >validation (eg Schemarama [4]). > >tia > >Damian Steer > >[1] rdfDB <http://www.guha.com/rdfdb/> >[2] NEOOM <http://www.nesstar.org/sdk/> >[3] MDL <http://www.charteris.com/mdl/> >[4] Schemarama <http://ilrt.org/discovery/2001/02/schemarama/> >-----BEGIN PGP SIGNATURE----- >Version: GnuPG v1.0.6 (GNU/Linux) >Comment: Processed by Mailcrypt 3.5.6 and Gnu Privacy Guard <http://www.gnupg.org/> iD8DBQE9NVk7AyLCB+mTtykRAukqAJ0UjoeKWJMOgS/KheaoV1tUyB6j3gCghK/V IuHUMv650H+5bqUmz4g0z6Q= =OdaN -----END PGP SIGNATURE-----
Received on Wednesday, 17 July 2002 15:22:54 UTC