[Sigia-l] OWL - Whoo, hoo!

[Adrian Howard]

On Monday, August 25, 2003, at 07:24  pm, Lars Marius Garshol wrote:
[snip]
> I think it helps to separate the pieces involved here: FOAF doesn't
> really have anything to do with OWL.

Erm... FOAF is an OWL ontology isn't it?

See <http://rdfweb.org/mt/foaflog/archives/000062.html>.

>  FOAF is an RDF application, and
> OWL is just an advanced schema/constraint language for RDF. What makes
> FOAF interesting is *not* OWL, but the fact that it uses RDF.

I wasn't trying to say the FOAF is interesting just because of OWL - 
but that FOAF is a good example of where using OWL, rather than a home 
grown solution, pays off.

OWL makes FOAF more interesting since it's one of the first vaguely 
popular public ontologies. Since the information is highly distributed 
and people are going to want to combine it with other "semantic web" 
content using an open standard is an obvious win.

> Personally, I think that for IA kind of things topic maps are a better
> fit than RDF, which is heavily oriented towards logic, and doesn't
> really give you much help with IA-type issues.

If it does become widely adopted I think it will make a difference to 
some IA areas.

For example, managing navigational structures that are created on the 
fly using inferences based on the OWL content from multiple sites looks 
like an interesting IA/usability challenge to me.

> | It's a nasty catch-22. Until other people start publishing OWL
> | documents then you don't gain much advantage in exporting OWL
> | yourself.
>
> Well, that's not really true. OWL is carefully designed to support
> logical inferencing. That's what it's for, and what it really does,
> and if you want logical inferencing OWL is very useful to you, even if
> you don't exchange it with anyone else. And if you *don't* want
> logical inferencing there's not really all that much you can use OWL
> for, except as a fancy schema language.

The point I was trying to make was that if you're in the market for an 
inference engine there are many systems that are more mature and 
simpler to implement than those based on OWL.

So, what would be my motivation to switching over to OWL if I have, for 
example, an existing Prolog based project that works well.

Integration with others seems the obvious reason. However, until there 
are people out there publishing OWL there is nothing to integrate with 
so OWL isn't worth the extra effort. Catch-22.

That's why I think FOAF is interesting from the OWL perspective since 
it offers something that other people can hang functionality off. I'm 
not 100% convinced by any means - but it seems the most likely starting 
point for wide acceptance of OWL that I've seen so far.

> | It's also going to be interesting to see how it impacts with data
> | protection issues. If OWL becomes popular then data aggregation
> | becomes *trivial*, which many people find objectionable.
>
> Uh, no, it does *not* become trivial. Not even nearly. There are quite
> a few data integration software companies selling software that has
> capabilities well beyond what OWL offers, but even they won't claim
> that their solutions are a panacea.

Your right. Trivial was an exaggeration... but if everybody does start 
adopting and publishing semantic content using standard protocols and 
formats it certainly becomes a *lot* easier than the existing networks 
of proprietary information.

If the semantic web does ever take off then we are going to see more 
and more information published in ways amenable to simple data 
aggregation. I think this will have unexpected consequences.

> | They're also the fact that as soon as something like OWL becomes
> | popular we will get the equivalent of search engine keyword-stuffers
> | trying to subvert the system in their own favour.
>
> For a semantic web-like vision that is definitely the case, but the
> identity-based data models like RDF are not only about being able to
> interchange information. They have many other useful properties that
> are helpful for information management even if you don't exchange data
> with others.
[snip]

Completely true. RDF is lovely  ;-)

Adrian