Re: Linkbases, Topic Maps,and RDF Knowledge Bases -- help me understand, please

[uche.ogbuji@fourthought.com]

> At 10:32 AM 4/6/01 -0600, uche.ogbuji@fourthought.com wrote:
> > > Even though XLink and RDF are targeted at different purposes, it's still a
> > > fair observation that XLink has a lot (not all) of the power of RDF.  For
> > > info on a suggested way to map XLink to RDF, see the note "Harvesting RDF
> > > Statements from XLinks":
> > >
> > >    http://www.w3.org/TR/xlink2rdf/
> >
> >I strongly disagree with this.  Just because basic RDF graphs can be
> >"harvested" from XLink doesn't mean that XLink has a lot of the power of
> >RDF.  RDF can be harvested from HTML as well, but that says nothing about
> >the power of HTML relative to RDF.
>
> I'm sorry you took offense.

Oh, no.  Strong disagreement, not offence.  I'm quite happy once again to
be discussing matters that interest me on this august forum.

> Maybe I should have separated my first and
> second sentences into two paragraphs.  The Note I referred to doesn't
> directly talk about the conceptual/data model similarities to XLink and
> RDF, but surely it's obvious that they exist, and the Note shows the most
> obvious way: usage of roles and arc roles in constructing assertions.  (The
> role stuff in XLink was explicitly added to allow for clean mapping to RDF
> statement semantics, and this choice was supported by and coordinated with
> some RDF folks.)
>
> I was not trying to say you could simply use XLink for metadata purposes or
> to cheapen RDF; rather I was pointing out that even though you can force
> them both into somewhat similar molds, they don't have the same purpose and
> so each has its place.  For example, XLink has "traversal" semantics that
> RDF doesn't, and RDF has (as I understand it) "bag" semantics that XLink
> doesn't.

Containers are one matter that I tend to group into the not-so-useful RDF
category (at least until the "_1", "_2" mess is sorted out), but these
could also be simulated in XLink using XPointer and a special container
element vocabulary.

Not that I'd consider it a good idea to do so.

> >With enough contrivance, one can, of course, get everything RDF can do in
> >XLink, for instance, one could use XPointers referring to XLink elements
> >to simulate reification, and one could concoct a schema system, but this
> >isn't a useful counter to RDF's facility.
> >
> >And then again there are aspects of RDF that are pretty much outside the
> >capacity of XLink, such as the flexibility of RDF's XML serialization
> >(which can be burden as much as boon, but is a fact).
> >
> >I actually don't think there's much overlap between RDF and XLink at all.
>
> I have to disagree, but I'm not really trying to get into a boxing match
> over which is "better";

Nor am I.  My point was not that RDF is better than XLink.  My
disagreement was with your assertion that XLink provides most of the power
of RDF.  This is not true; nor is it true that RDF provides much of the
power of XLink.  They both have quite different strengths.  I personally
use each of them heavily, and I wouldn't dream of using RDF to implement
content links, which is what XLink does very effectively.

> I think they both have their strengths and use
> cases, and the RDF-knowledgeable people with whom I've worked in doing
> XLink stuff have agreed.

I agree as well.  I think you misunderstood me, but it seem I might have
misunderstood you as well, to a point.


-- 
Uche Ogbuji                               Principal Consultant
uche.ogbuji@fourthought.com               +1 303 583 9900 x 101
Fourthought, Inc.                         http://Fourthought.com
4735 East Walnut St, Ste. C, Boulder, CO 80301-2537, USA
Software-engineering, knowledge-management, XML, CORBA, Linux, Python