OWL and RSS 1.0 [was: Re: RSS 1.1]

Leigh Dodds wrote:

> I'm disappointed that the RSS 1.1 RDF/OWL schema doesn't make any 
> reference to the RSS 1.0 classes and properties. Wouldn't this have 
> been a nice demonstration of RDF format evolution?

Christopher has already dealt with this in a practical sense, but I want
here to deal with the underlying rationale. Sadly, any RSS 1.0 feed that
uses images incorporates <image> as both a nodeElement and a propElt,
which is mandated by the specification. In other words, rss:image is
used as both a class and a property. These are disjoint in OWL DL:

[[[
LVI, IOT, IOC, IDC, IOOP, IODP, IOAP, IOXP, IL, and IX are all pairwise
disjoint.
]]] - http://www.w3.org/TR/owl-semantics/rdfs.html#5.4

But moreover they're disjoint in common sense; the RSS 1.0 schema is
broken in this regard. So this is the main issue: that were we to
provide relationships to RSS 1.0, we'd be providing them to an
inconsistent schema. You don't define a bugfix release in terms of the
specification whose bugs you're fixing!

A related issue is the use of further constructs in the RSS 1.0 schema
(per [1]) that make it OWL Full, i.e. use of rdfs:Class:

[[[
NOTE: owl:Class is defined as a subclass of rdfs:Class. The rationale
for having a separate OWL class construct lies in the restrictions on
OWL DL (and thus also on OWL Lite), which imply that not all RDFS
classes are legal OWL DL classes. In OWL Full these restrictions do not
exist and therefore owl:Class and rdfs:Class are equivalent in OWL Full.
]]] - http://www.w3.org/TR/owl-ref/#ClassDescription

Note that the RSS 1.1 ontology is in OWL Full, but that it only uses an
OWL Full constuct in a two omittable places. After conversations with
Joe Geldart on the matter, we decided that these constructs (the use of
rdf:List, and one rdfs:Class elsewhere) was a beneficial restriction
that could easily be omitted for OWL DL and OWL Lite tools.

So though the RSS 1.0 rdfs:Class issue isn't a clincher, if we were to
declare RSS 1.1 terms as being equivalent to RSS 1.0 terms, we'd be
binding ourselves ever more tightly to OWL Full, which is inconvenient.
On the other hand, were we to define them as subClasses, it's possible
that we'd find that RSS 1.0 class instances are instances of that part
of rdfs:Class which is in the owl:complementOf owl:Class, so once again
there's a possibility that we'd be incompatible with OWL--though we'd
have to investigate this more carefully.

Note that my use of qualified cardinality constraints, a la DAML+OIL,
has already been raised as a point which isn't clearly covered by the
OWL Reference Guide: it suggests that qualified cardinalities are
unacceptable, but Sean Bechhofer's OWL validator passes it, so we're
unsure. We'll be changing them to unqualified cardinality constraints
anyway since this doesn't change the intended semantic but does, as far
as I know, make it OWL DL compatible.

We'll likely summarize these issues properly somewhere else, perhaps in
the next draft of the specification, and ask www-rdf-logic or the
nearest available OWL expert for more feedback.

[1] http://web.resource.org/rss/1.0/schema.rdf

-- 
Sean B. Palmer, http://inamidst.com/sbp/


 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/rss-dev/

<*> To unsubscribe from this group, send an email to:
    [email protected]

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/