model vs. triple

Hi,
I've been working with RDF for about three months, and am working on
an application. A colleague of mine and I disagree about the most
appropriate way to deal with the data. He thinks the data should
be dealt with at the model level, that a model plays the central role
when working with the data. Models are what get inserted into and
deleted from databases, and models are what is returned from
searches. I think the data should be dealt with at the triple level,
at least for our application. I think that a model is a useful way
to talk about groups of triples, but not a necessary way. Triples
get inserted, updated and deleted, and searches return triples.
Anything said about a model is secondary to the triples.
  
For instance, I have a triple that looks something like this:
{dcq:modified, "http://website.com/page.html", "2001-05-15"}
I want to be able to update that triple as needed. Just the
triple. If I'm working at the model level, I have to update the
whole model, because of a tiny change.

So, I'd like to know if anyone here thinks that I'm on the wrong track.
Is it wrong, or likely to cause problems, to take the triple-centric
path. I think that while both are acceptable ways to work with the data,
the triple-centric view allows for more powerful manipulation of
the data.

Devon Smith
smithde@oclc.org

Received on Tuesday, 29 May 2001 09:03:35 UTC