Re: Provenance in RDF

> There are no assumptions in our system about whether the fourth member of
> the quad (fourth in addition to the pred, subj, and obj that is) is a
> statement id or a context id. In the site I referenced in my last email we
> use it as a statement id (i.e. each statement has a unique id).
> Our query language doesn't require the use of the quads but does allow its
> use. Internally, our system is quad based. If the "context" isn't specified
> in a query (i.e. a triple is used instead of a quad), we assume a  null
> context (a particular well-known uri). A triple-based query only sees
> statements in the null context (contrast that with conventional triple
> refiication where queries are clogged with the machinary of reification -
> the reifying statements themselves).

4RDF has a similar concept, which we call "domains".  A domain is basically a 
URI associated with a collection of statements.  Most naturally, this is used 
to "keep track" of the URI from which a particular set of ststements was 
deserialized (i.e. an RDF/XML source document).

But domains can also be used to manage context, and to record provenance.

We don't support domains in Versa because we wish it to be independent on 
4RDF, which happens to have its first implementation.

However, we will probably add Versa extension functions for 4RDF to 
accommodate domain restrictions.

Uche Ogbuji                               Principal Consultant               +1 303 583 9900 x 101
Fourthought, Inc.                
4735 East Walnut St, Boulder, CO 80301-2537, USA
XML strategy, XML tools (, knowledge management

Received on Friday, 1 March 2002 14:40:07 UTC