[rdfweb-dev] Describing FOAF using OWL: enumerating allowed

Dan Brickley danbri at w3.org
Sun Jun 22 17:44:49 UTC 2003

* Morten Frederiksen <mof-rdf at mfd-consult.dk> [2003-06-22 19:27+0200]
> On Sunday 22 June 2003 19:08, Danny Ayers wrote:
> > I'm not sure I understand, isn't this just moving the problem (controlled
> > list of literals) to the ENTP-person class? Where would you express the
> > Alt-like options ENTP | INTP | (whatever else there is)?
> This is just an addition to the controlled list, it's not an alternative.
> (Although I guess one could possibly define the 16 classes, relate them to
> each other somehow, and state that a person is in one class only.)
> The point was that using literals did not necessarily mean that it wasn't
> possible to make statements about the "literal values", just like it can be
> done if URIs were used.

In fact, this is much better for many purposes, since we're making claims 
about what those literal properties _tell us_, and not merely providing 
annotations on the values themselves.

You could claim (falsely or whatever) that all Google employees are 
americans, or that everyone who went to my school has a nearestAirport of 
BRS or many other unwise assertions. OWL is fantastically expressive in 
that regard. In other ways, eg. representing the rules for something like
x being the foaf:uncle of y, it is somewhat limited.

Figuring out just what it can't do, and how it relates to RDF query languages,
rule languages and such like, will require some real world experience with 
real world data. Hence this thread :)


More information about the foaf-dev mailing list