Re: DAML Level of Effort for FY03-FY05

I do think imports would be better suited in RDF; I think it is a
fundamental building block of the Semantic Web, and without it, RDF is
just another data modeling language. Hopefully, WebOnt will standardize
the notion and it will eventually get pushed down into RDF.

For a formal treatment of how I see imports working, you can see Chapter
3 of my thesis [1] (which Didier mentioned earlier). This chapter builds
a logical foundation for the Semantic Web based on first-order logic. It
explains why ontologies are necessary and why resources need to be able
to import ontologies (in the thesis, this relation is called the
commitment function). I also discuss ontologies importing other
ontologies (called ontology extension in the paper) and an initial
mechanism for handling the evolution of ontologies over time. I
introduce a notion of perspectives that allows us to define standard
ways of combining resources in the presence of multiple inter-related
ontologies. Note, this chapter is not for the casual reader: it is
math-heavy (the rest of the thesis is much more readable). I've been
planning on writing a journal article on the topic which clarifies and
expands on the notions, but haven't gotten around to it yet. Hopefully,
this summer...

Jeff

[1] http://www.cse.lehigh.edu/~heflin/pubs/#heflin-thesis

Didier VILLEVALOIS wrote:
> 
> Hi Dan,
> 
> I do agree the meaning of the import property is not 'yet' clear.
> I think the *perspective* concept that Jeff Heflin introduced in his thesis
> is a good representation of views over the semantic web.
> 
> In fact, i can't see any reason to use something else than the namespace
> declaration
> just to refer to a term. (and that's what i do in my software)
> For me, *importation* would be more a kind of aknowledging of the model
> theory (meaning)
> implied by how this term is defined.
> 
> What i suggest anyway is that we need a super-property for all properties
> that define links
> between documents. Maybe you'll find that rdfs:seeAlso or rdfs:isDefinedBy
> is sufficient but i need
> to refine their semantic.
> 
> Didier.
> 
> [Dan]
> > All the while 'imports' isn't in the core of RDF, people can
> > reasonably
> > claim they didn't know what it means, that it is just some utility
> > vocabulary. If we put it in the main specs, we'd need to be a
> > lot clearer
> > on what conclusions 'imports' allows us to draw about a
> > document and its
> > author/publisher.

Received on Tuesday, 23 April 2002 16:04:06 UTC