xml-dev - RE: RDF, the "semantic web", and the nadir of AI (was RE: Realist icprop

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

If the namespace ID for the schema is bound to the 
component which implements it via some GUID, then 
the semantics are bound to the data.  Data binding 
is a straightforward issue for small sets.  If the 
namespace by value or reference attempts to scope 
too many terms, then the associations get weaker 
and the components become fragile.  MIME works sorta.

Somewhere in the back of all this, I suspect, is someone 
who thinks without sufficient experience, that we 
really will have a finite set of schemas which 
will be exhaustive with respect to the domain of 
the services.  That is HTML thinking.  For the reasons 
Gavin pointed out, this typically fails.  And by the 
way, CALS tried it too with MIL-D-28001.  No size fits all.

Now, if a non-exhaustive set is sufficient because 
it is recognized that the value chains of services 
are semi-autonomous (operate with some ecotonal 
overlap but otherwise, are discrete), then the 
schema-based systems cohere reasonably well.  This 
is how database systems that interoperate through 
common APIs, commodity transport protocols, and 
portable validatible data formats work today.  
What XML and schemas bring to the party is a 
somewhat cheaper means to negotiate the bridging 
documents, once called intermediate file formats. 

So in effect, some progress (commom means) and 
therefore, more efficiency in creating the bridges.  
The Desperate PERL Hacker is able to do the job 
for which we once needed a C++ network guru.

Len Bullard
Intergraph Public Safety
clbullar@ingr.com
http://www.mp3.com/LenBullard

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h


-----Original Message-----
From: Rick JELLIFFE [mailto:ricko@geotempo.com]
Sent: Wednesday, October 18, 2000 7:51 AM
To: Robin Berjon; xml-dev@lists.xml.org
Subject: Re: RDF, the "semantic web", and the nadir of AI (was RE:
Realistic proposals to the W3C?)


Robin Berjon wrote:
> 
> At 03:40 18/10/2000 -0400, Gavin Thomas Nicol wrote:
> >> I don't know whether or not the semantic web will
> >> succeed, but the idea does make some sense.
> >
> >Please explain it to us all... I've never really grokked it,
> >and I used to work in AI...
> 
> I'm not sure it *has* to be looked at from an AI perspective.  

If we can give TBL the benefit of the doubt, I think it is possible that
the "semantic web" means--at least--something.  At a minimum, surely it
is a web of atoms of information each component of which can be
universally addressed, and where the arcs between each node (or the node
itself) has some label, and that if one can trace back along these arcs
(including schemas to bring the labels into the web too) to well-known
datums (IYKWIM) then one can do more or less useful things with that
web. 

This is not AI, this is just a big fat database. Is dog has a collar; a
dog is an animal; an animal can have a name; a dog can have a collar, a
collar can have a name tag; a nametag can have a name: start from the
dog and do a search of everything connected to it to try to find the
name.  AI comes into the heuristics in navigating around a database of
information.

The question is how much the technology we are building actually
promotes that: an XML Schema is not "semantic" in the kind of sense
above--it gives information for types not meanings or properties. This
is one objection to identifying namespaces and schemas too much: it
actively prevents a semantic web (in that sense.)

Cheers
Rick