The semantic web

Hi all,

"Practical Knowledge Representation for the Web" 
(http://www.cs.vu.nl/~frankh/postscript/IJCAI99-III.html#Calvanese:98AAAI)

because the paper Frank and I have been written did pop up
again in the discussions I would like to make some general
comments on it.

1. We wrote the paper at a point in time where we thought RDF
aims on more that it actually does. We thought that it is
meant as a web-based modelling language for meta data. 
Therefore, we critized RDF for lacking many features
one would expect for a suitable modeling language.
Meanwhile we learnt that RDF is "just" a syntax for
writting down triples (very similar to a representation
formalism covering only binary predicates). Clearly
one could critique RDF for some of its design decisions,
hard to read syntax, and explanation. However, having
such a standard for writting down triples is a good
thing anyway. 

2. RDFS starts to be on a more interesting level because 
it introduces additional modeling primitives coming
closer to a language that one would actually use for
modeling meta data that are a bit more complex and
structured than simple triple sets. Still, the
primitives offered by RDFS are somehow limited and
these choices are not really motivated. At least we
fail to get these rationales from the scheme specification.

3. What we have done (in cooperation with many other
research groups and industrials) was to define an Ontology
Inference Layer OIL on top of RDFS (i.e., we define
OIL as an extension of the name space of RDFS).
More details can be found at 

	http://www.ontoknowledge.org/oil.

The main features of OIL are:

	- providing frame-based modelling primitives,
	because they are rather intiutive to most
	users;

	- defining a formal semantics and reasoning
	support based on Description Logics and the
	FaCT reasoning system;

	- defining a syntax based on RDF and RDFS
	(i.e., we define the extensions of OIL
	in RDFS). For people who do not want to
	follow the RDF(S) bandwaggon we also provide
	XML schema definitions of OIL.

In consequence, we define an inference service layer
on top of things like RDF and RDFS. Therefore, I think 
we closely follow the vision shared by many people
(including Tim). Our early critique on RDF was based
on the missunderstanding that we thought that already RDF 
should provide such a service.


Greetings,


Dieter

----

Dieter Fensel
Division of Mathematics & Computer Science,
Vrije Universiteit Amsterdam, 
De Boelelaan 1081a, 1081 HV Amsterdam, NL
The Netherlands
Room number U3.25.  
Tel.: +31-(0)20-444 7739, 
Fax and Answering machine: +31-(0)20-872 27 22
Mobil phone: +31-(0)6-51850619
Email: dieter@cs.vu.nl
http://www.cs.vu.nl/~dieter
Privat
Liendenhof 64, NL-1108 HB Amsterdam, The Netherlands. 
Tel.: +31-(0)20-365 52 60.

Received on Wednesday, 12 April 2000 18:53:42 UTC