[topicmapmail] Inference rules

[Murray Altheim]

Stefan Henke wrote:
> Hi,
> I have a very basic question about topic maps. Which role do inference rules
> play in the world of topic maps? I think in the field of ontologies, they
> play a quite important role. In the last time, I had a look at the tolog
> query language. It allows to define inference rules which are used for the
> query. But the rules are not getting part of the topic map itself. So, they
> are not used in other queries. Are there any plans to build in some concepts
> for inference rules in topic maps?

Inference rules are simply rules that can be interpreted by an
inference engine, based on whatever ontological commitments are
stated by those rules. So if a rule states the logical AND or OR,
an association that represented that rule could be reasoned upon
in the same way as any logical AND or OR statement. By extension,
any statement (simple or complex, in whatever form of logic)
expressed within the graph of a Topic Map can be reasoned upon.
Topic Maps are simply (in this sense) another means of making
such statements.

There is no reason to attempt to build inference rules into the
Topic Map standard, as they are a natural consequence of whatever
semantics are being expressed. Because the terminology of logic
requires a high degree of precision, and because there are as
many applications of logic as there are domains of application,
it doesn't make sense to confine, say, the definition of "class"
within the Topic Map standard to one of the many possible
definitions. If an application requires a specific set of
semantics, it can define those semantics (say, as a set of PSIs,
or simply as Topics), state the concomitant inference rules, and
perform whatever inferencing is desired based on those rules.

For example, any Topic Map application that takes advantage of
the XTM 1.0 class-instance and superclass-subclass associations
is already doing inferencing, and certainly there are a number of
applications that have gone much farther than this.

As another example, if someone wanted to implement a Topic Map
that expressed the same semantics as John McCarthy's original
Situation Calculus, an inferencing engine could reason upon the
resulting Topic Map in *exactly* the same way as it could the
same expressions stated in the textual formalism used in the
Situation Calculus by McCarthy, e.g.:

   ¬Holds(On(R), S1) ^ Holds(On(Sw), S1).

There'd be some equivalant expression of this in XTM. It's just
a difference of the language chosen to express those semantics.
This is similar to OWL as an expression of Description Logics in
the RDF syntax. This all could be done in GXL too. If you consider
that R.V. Guha was the one who brought RDF to the W3C, and look
at his background, this all makes pretty good sense. In order to
accomplish any of his requirements, he needed some graph theory.
Well, we have some graph theory.

But I see little reason to standardize these things -- publish,
yes -- but not standardize.

I hope this answers your question satisfactorily.

Murray

......................................................................
Murray Altheim                    http://kmi.open.ac.uk/people/murray/
Knowledge Media Institute
The Open University, Milton Keynes, Bucks, MK7 6AA, UK               .

   The Resume of George W. Bush
   http://www.buzzflash.com/contributors/03/04/23_resume.html

                    The sleep of reason brings forth monsters -- Goya