W3C home > Mailing lists > Public > www-rdf-interest@w3.org > May 2002

RE: Taking an axe RDF in XML? (no thank you)

From: Assini, Pasqualino <titto@essex.ac.uk>
Date: Fri, 24 May 2002 12:15:28 +0100
Message-ID: <7AC902A40BEDD411A3A800D0B7847B6676B38E@sernt14.essex.ac.uk>
To: "'kifer@cs.sunysb.edu'" <kifer@cs.sunysb.edu>
Cc: www-rdf-interest@w3.org
Hi Michael,

> A more convincing use of RDF (and by "RDF" I mean a broader 
> language for
> the semantic web) could be if your company had something like 
> UDDI and WSDL
> expressed in RDF. 

Indeed we have :-)

We have a simple RDF model to specify methods (services if you prefer). 

Being expressed in RDF it can be both way simpler (essentially 2 classes)
and more powerful (support inheritance, methods can have methods and
properties, method invocations are objects in their own right, etc) than
WSDL.

We have ways of searching for objects/services but this is just an add-on
(just as google is an add-on to the web, fantastically useful but not part
of the basic architecture).

The system is composed by a set of objects that "live" at a certain URL on
the web (just as any other web resource) and self-describe themselves in RDF
(if you access their URL you get back their description in RDF).

The RDF description naturally mentions the class of the object and at the
class URL you will find the RDF Schema of the class, its properties and
methods.

That's all there is to it.

You can check one of these objects at
http://nesstar2.essex.ac.uk/neoomtest/servlet/NEOOM2EJB/obj/cServer/ESSEXTES
T.

This is an object that represent a server and provide a simple
administrative interface.

As the RDF description is pretty dull you might be better off looking at it
through a simple object browser:  

http://nesstar2.essex.ac.uk/browserFRAMES=true&url=http://nesstar2.essex.ac.
uk/neoomtest/servlet/NEOOM2EJB/obj/cServer/ESSEXTEST


> One could query the UDDI/RDF directory and determine (using 
> some knowledge
> and criteria) which of the many services to invoke. Then 
> check the WSDL/RDF
> piece and figure out how to invoke the service. If it could 
> be demonstrated
> that such a thing can be done very succinctly in the expanded 
> RDF, I can
> imagine that this might be an incentive.

The object browser does just that.

It will retrieve the instance description in RDF, find out the object class,
get the class description and use it to built a simple web interface that
allows you to inspect all the properties and apply all the methods that the
object support (more about it at http://www.nesstar.org/sdk/).


 
>     AP> I am not thinking of natural language processing, 
> just a simple
>     AP> syntax that 
>     AP> can be parsed with ordinary tools but that could be 
> interpreted as
>     AP> a simple 
>     AP> form of human language.
> 
>     AP> Something like:
> 
>     AP> http://www.yahoo.com is a HomePage and its author is 
> 'John Smith'.
> 
>     AP> Reification:
>     AP> 'John Smith' says "http://www.yahoo.com is a HomePage".
> 
> There is a (big) difference between what a programmer (or a knowledge
> engineer) needs and what a "naive" user can swallow. For a naive user
> something like the above can be provided through a GUI -- a 
> dumbed down 
> fact language for representing facts. It is not a big deal to 
> even design
> an interface where such a user would be able to specify simple rules.

You are right.

But a simple textual syntax would be, just as you say:

"an interface where such a user would be able to specify simple rules."

or maybe just simple facts.

It's true that you can have visual tools instead and I would expect most
users to partecipate in the semantic web by the intermediation of such tools
but the fact is that writing down something is one of the most efficient and
more familiar way of expressing information and so it cannot be so easily
dismissed.   


> For a programmer/knowledge engineer, however, you don't want such an
> interface.  This is one of the mistakes that was made by the 
> designers of
> SQL. (Jim Gray once said, "I designed it and I apologize". He wasn't
> completely serious, but it wasn't a pure joke either.)

I think that you are 99.9% right.

But just let me elaborate briefly on the remaining 0.1%.

I am fascinated by natural language and by how we might extract and
formalize bits of it to use in the semantic web.

After all the bread and butter of knowledge engineers ("is-a", "part-of",..)
are just bits of semantic stolen from natural language.

There is much more to learn from it and trying to express formalized
knowledge in a pseudo-natural language form might help focus our mind on
this fact and help us take advantage of it.

Best 


     titto
 
-------------------------
Pasqualino "Titto" Assini - Nesstar Ltd
John Tabor Building - University of Essex
Colchester, Essex  - CO4 3SQ  - United Kingdom
email: titto@nesstar.com <mailto:titto@nesstar.com>  personal email:
titto@kamus.it <mailto:titto@kamus.it> 
Received on Friday, 24 May 2002 07:17:48 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:54 GMT