[sc34wg3] TR: comment - RDFTM: Survey of Interoperability Proposals

Robert Barta sc34wg3@isotopicmaps.org
Sat, 12 Mar 2005 06:16:32 +1000


On Fri, Mar 11, 2005 at 03:36:00PM +0100, Lars Marius Garshol wrote:
> | It is the field of mathematical logic (including set theory and
> | others) that provides the tools by which we may create model
> | theories.
> 
> Correct. \Tau doesn't use mathematical logic, nor model theory,
> however. Hence my suspicion that \Tau wouldn't satisfy
> Patel-Schneider's requirements.

The models themselves, so what the potential structures of maps are,
these are defined via sets. There is no logic involved; maps - by
themselves - do not mean anything.

The _operators_ on these structures are defined _using_ first order
logic. But the operators (moving through the map) themselves are not
regarded as logic. I think.

Path expressions like "give me all association in the map m which are
an instance of 'is-part-of'"

   m [ \pi -> class = is-part-of ]

are evaluated on the model. Either they return something (TRUE) or not
(FALSE). Path expressions are logic formulas and the path language is
a logic _tailored_ for TMs.

So, this looks like classical model theory to me (but I am not an
expert).

No doubt, it would be interesting to analyze how path expressions are
related to DL, CL, ... For some ideas, see the last section in

   http://topicmaps.it.bond.edu.au/docs/30/toc

\rho