Re: semantics for SWOL : initial message

         "The Problem with Semantics in DAML+OIL"

Dear Peter,

excellent work! Honestly it took me quite a while to understand the problem you
encountered. I think you make a very good point here but I am not sure whether
I agree with the solution you propose. In any case, there need to be an action
when things like a disjoint statement becomes a relation in the model theory.
This would get even worse when in a latter stage more expressive languages
with richer axioms will be defined. I think there are basically two ways to go:

         (1) referring to RDF as a syntactical layer
         (2) referring to RDF as a semantic layer

If I understand you right, then you proposing way no (2) whereas I would prefer
way no (1). I will try to motivate my choice.

Basically in good old OIL days we used RDF as a syntax and did not need to
worry about any entailment in RDF because it was not defined. We also defined
OILcore in a way that excluded all the bizzare meta model stuff I never under-
stood (and if I understood it I forgot it five minutes later again). 
Therefore, we
always had in mind that an RDF agent can make additional conclusions that
an OILcore agent is not able to draw because he does not make use of the RDF
features that allows you to define the language in itself. Also an OILcore
agent can make conclusions that an RDF agent is not aware of because
the latter does not understand semantics of things like "disjoint".

I would see entailment defined in RDF as a means to reason about the syntax
of a language or Ontology (with RDF reasoning you can make a distinction
whether a disjoint statement is defined explicitly or whether it is a logical
conclusion of other logical statements) and entailment defined in WOL as
a means to reason about the semantics of an Ontology. I think RDF should
NOT viewed as a basic Ontology language (and I have to update my own
slides on this) but as a syntactical mechanism to define Ontology Languages
in it. From the standpoint of an Ontology language, RDF has many strange
features and thanks to Pat they are made more explicit now. On the other
hand, they may make sense if you want to use RDF as a mechanism to
define Ontology languages in it.

Finally from a strategic point of view, I do not think that we can understand
the tower of languages for the semantic web in a strict monotonic way that
each level entails a super set of facts compared to a layer below. I would
not be surprised if each layer enables different types of entailment where
some are overlapping and others not. Spoken in a nutshell:

         1. I would recommend to stay syntactically as close as possible to
         RDF to allow an RDF agent to understand as much as possible (in
         his RDF style on looking at the world) from WOL.

         2. I would not recommend to define the model theoretical semantics
         of WOL as an extension of RDF because RDF serves a different
         purpose than an Ontology Language. I would recommend to define
         it independent from it. I always think that [Kifer et al., 1995]
         provides a nice pattern on how to define a semantics for a simple
         Ontology Language and I used in this in my PhD as a pattern.
         Basically this also implies that RDFS is not necessarily viewed
         as a basic Ontology Language but a schema language for RDF
         (like XML schema is not necessarily viewed as a basic Ontology
         Language but a schema language for XML).

         3. A third option would be to re-define RDF and RDFS in a way that
         they become suitable as a basic Ontology language. But this
         does not look likely to happen and it may be even the wrong way
         to go for the RDF people.

What looks like a technical question may be quite important for how to
organize the semantic web. If we try to tie these languages to closely they
may either no longer fulfill their purpose or you loose too much. I view 
the fact
that an RDF reasoner can make conclusions beyond the scope of a WOL
reasoner as a feature and not as a bug.


Greetings (and thanks for the great work you are doing. It helped me a lot
in understanding things better),

Dieter


[Kifer et al., 1995]
         M. Kifer, G. Lausen, and J. Wu: Logical Foundations of 
Object-Oriented
         and Frame-Based Languages, Journal of the ACM, 42, 1995.




----------------------------------------------------------------
Dieter Fensel
Division of Mathematics & Computer Science,
Vrije Universiteit Amsterdam,
De Boelelaan 1081a, 1081 HV Amsterdam, NL
The Netherlands
Tel. (mobil): +31-(0)6-51850619,
Fax and Answering machine: +31-(0)84-872 27 22
Email: dieter@cs.vu.nl
ICQ #132755538
http://www.google.com/search?q=dieter or http://www.fensel.com
Privat: Liendenhof 64, NL-1108 HB Amsterdam, The Netherlands.
Tel.: +31-(0)20-365 52 60.
----------------------------------------------------------------
The information transmitted is intended only for the person or entity to which
it is addressed and may contain confidential and/or privileged material. Any
review, retransmission, dissemination or other use of, or taking of any action
in reliance upon, this information by persons or entities other than the
intended recipient is prohibited. If you received this in error, please
contact the sender and delete the material from any computer.