John,
Relevant to this thread because it is relevant to the issue of "content":
Concerning your opinion: (01)
[JS] > It's impossible to have
> some kind of abstract, universal, task-independent interoperability. (02)
[PC] I am not convinced of this, because I believe that there is a finite
number of conceptual primitives, and that interoperability depends on
sharing the same conceptual primitives, with which the non-primitive terms
can be defined. **IF** there are a finite number of conceptual primitives,
**THEN** any two systems using the same conceptual primitive representations
can interoperate accurately by sharing their definitions as well as the data
they want to exchange. (03)
This does not, however, mean that such a process is the most
**efficient** means to interoperability in any one case- I am concerned here
only with the theoretical basis of broad and accurate interoperability. For
any two (or small number) of given systems, specially devised ad-hoc data
conversion methods may be (much) more efficient than going through several
levels of definitions grounding in the primitives. (04)
[PC] >
> Even human beings, with the full flexibility of human intelligence,
> are only capable of collaborating successfully on a limited number
> of tasks without going through a lengthy period of training.
> (05)
[PC] Yes, agreed, but the "training" period for computers that already can
process the necessary set of conceptual primitive representations can be
milliseconds- updating the ontology to include new terms defined by the
system sending data. (06)
[JS] >
> AW> To get a measure of how far there is still to go on this journey,
> > here's a real world problem description that serves as a nice
> > example of an interoperability requirement...
>
> The financial system is an excellent example. Double-entry bookkeeping
> and the basics of banking services were developed during the Italian
> Renaissance. The terminology for banking services is fairly universal.
> Banks have been able to do electronic funds transfers and other kinds
> of interactions for many decades.
>
> But... (and this is a big *BUT*) whenever two banks merge (which they
> often do), they *never* merge their databases. Instead, they always
> adopt one of two procedures:
>
> 1. They continue to run the software and databases from *both* banks
> for the accounts that came from those banks.
>
> 2. They close the accounts of one of the parent banks, transfer the
> funds to the other bank, and reopen new accounts in that bank.
>
> Usually, they adopt procedure #1 in the initial stage of the merger
> and gradually use repeated applications of procedure #2 to convert
> accounts. Nothing that we have ever discussed in this forum will
> make any difference in that merger process.
> (07)
[PC] Yes, this is a good example of the practical issues that impel toward
special ad-hoc methods for interoperability among a small number of systems.
I expect that such a method will probably always be preferred when the
systems that need to interoperate can be controlled by a single general
manager. But the **general** question of whether a very broadly applicable
(if not actually "universal") semantic interoperability - among many systems
that cannot adjust to each other directly - can be supported by the
widespread use of a common set of semantic primitives does not seem to
inherently impossible - for which reason I disagree with your emphatic
general assertion. I think this is an empirical question that can only be
resolved by actual testing, in the engineering mode: build such a system and
see if it works. (08)
Of course, there are many questions that need to be resolved, including
just what qualifies as a "primitive" (for which I have some thoughts that
can be discussed in some other thread). I think the theoretical question of
whether the number of primitives is limited is of great importance to the
future utility of ontologies, and should be one of the topics for
substantive research within the community of ontologists concerned with
improving the performance and applicability to practical problems of
ontologies. (09)
Pat (010)
Patrick Cassidy
MICRA, Inc.
908-561-3416
cell: 908-565-4053
cassidy@xxxxxxxxx (011)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (012)
|