[tmcl-wg] Any TMCL dependencies: OWL Maybe?

[Robert Barta]

On Mon, Mar 24, 2003 at 10:24:13AM +0900, Mary Nishikawa wrote:
> Please take a look at section  4.1.2 Converting OWL in Lars Marius 
> Garshol's paper, "Living with topic maps and RDF."
> 
> http://www.ontopia.net/topicmaps/materials/tmrdf.html#sect-converting-owl
> 
> This too:
> 
> 4.1.3. Conclusions
> The obvious conclusion is that converting the constraints in an RDF Schema 
> or an OWL ontology to the future TMCL should not be too difficult, provided 
> an RDF to topic maps mapping is provided.

I think the RDF <-> TM mapping has per-se nothing to do with the OWL
<-> TMCL mapping.

In the case of RDF <-> TM mapping we have two 'documents' which are
instances of their respective data models. If we have a clear
understanding what "the meaning of construct A in RDF" is then we can
translate it into something "equivalent" with construct B in TM and
vice versa.

In case of OWL we are talking about a "logic", in case of TMCL we
probably do as well. Whether the two logics are equivalent depends on
the expressiveness (horn-clause, first-order, ....). Here we would
have to map the models in a model-theoretic sense.

According to

   http://www.w3.org/TR/owl-semantics/

they are using Description Logic (DL) which is a subset of predicate
logic:

   http://www.ida.liu.se/labs/iislab/people/patla/DL/index.html

   "The basic building blocks are concepts, roles and
   individuals. Concepts describe the common properties of a collection
   of individuals and can be considered as unary predicates which are
   interpreted as sets of objects. Roles are interpreted as binary
   relations between objects. Each description logic defines also a
   number of language constructs (such as intersection, union, role
   quantification, etc.) that can be used to define new concepts and
   roles. The main reasoning tasks are classification and satisfiability,
   subsumption and instance checking."

Whether our TMCL will be doing the same but - instead off using RDFish
concepts using topics, instead of RDF statements TM associations - or
not will have to be seen.

I personally think that DL is a good mix of expressiveness and
reasoning feasibility. Predicate logic is nice, but might be an
overkill.  During the use cases I have not found too many constraints
which need a stronger logic. It all depends what you understand as
"ontology" and how much application specific constraints you want to
take on-board, of course.

> It also appears that the parts of OWL that go beyond simple
> constraints can be reused directly in topic maps for the most
> part. For this reason it seems best if TMCL is created to only
> support constraints, and that rather than define a topic map
> ontology language OWL is reused as it stands within the topic map
> standards family.

I am not sure what Lars is here referring to. I guess he means
operators to build classes from existing classes. If so, I fully agree.

> Will the TMCL depend on XML Schema Part 2, or OWL or something else?

The "Data Types"? To a certain extent maybe. There is also a "Type
System" component in XML schema. I wonder whether this cannot be more
naturally be modelled in TM. Maybe Appendix C,D is most relevant here?

\rho