Martin and Aldo, (01)
I strongly agree with your general comments on the relationships between
theoretical ontologies and practical applications. I'd just like to
make a few comments on some of the points. (02)
MH
> Web ontologies sit somewhere between (many) people and machines,
> so their type system must try to balance out conflicting requirements,
> for instance what I called the "degree of disambiguation / discriminatory
> value of a type" vs. the reliability of type membership. (03)
My only point of disagreement is that I don't believe that there are
*any* ontologies that don't have to face such conflicts. Even in pure
mathematics and theoretical physics, theoreticians continue to use the
same words -- number, algebra, mass, force, energy... -- but they
keep changing the definitions. (04)
In computer science, just look at the definitions of 'file' in every OS
for IBM mainframes, the many versions of Unix and Unix-like systems,
MS Windows, Apple's OS, etc. Every patch to any OS changes some
definitions somewhere in the system. (05)
MH
> the reliability of type membership information) can suffer from more
> subtle distinctions in the type system, because human users who are
> publishing data, either by explicit annotation or by writing mappings
> to legacy data structures, may not be able to understand and/or apply
> the conceptual distinctions. (06)
Even professors, researchers, and their graduate students make such
mistakes. For computer systems, even superprogrammers make mistakes.
Nobody can keep up with all the details and changes to the systems
they use -- or the systems their programs have to interoperate with. (07)
MH
> So we cannot take it for granted that an ontologically better type system
> with cleaner distinctions enhances interoperability and reuse of data. (08)
Yes. Never assume that the users have read, understood, remembered,
and observed every distinction. In fact, you can't even assume that
the people who wrote the definitions will use them correctly in every
instance. And even if you have computers that mechanically check
every use, you can never be sure that all of them have upgraded to
the same version of the same OS with exactly the same patches. (09)
This point, by the way, undermines the assumption that universal
URIs or IRIs will magically solve the problems. Readable names
are more reliable than pointers to definitions that nobody reads. (010)
MH
> By the way, I am not saying that this holds for all ontologies. (011)
I'm not as optimistic as you are. That's why I would drop the
qualifier 'web' in front of 'ontologies'. (012)
AG
> There is an interesting psychological study on the value of ambiguity
> (especially when it is associated with functional relatedness). (013)
Thanks for the reference. (014)
From http://www.sciencedirect.com/science/article/pii/S0010027711002496
> We present a general information-theoretic argument that all efficient
> communication systems will be ambiguous, assuming that context is informative
> about meaning. We also argue that ambiguity allows for greater ease of
> processing by permitting efficient linguistic units to be re-used. (015)
To be learnable, every vocabulary must be finite. But no finite
vocabulary can precisely represent all the variations of science,
technology, business, arts, and life. (016)
Any attempt to eliminate ambiguity is not only futile, it's *worse*
than useless -- primarily because some starry-eyed idealists might
actually believe that the terms are unambiguous. (017)
AG
> In general, I'd say that ontology engineering should shift towards
> an empirical approach that validates and motivates the application
> of formal theories, logic, and philosophy in practical applications. (018)
Yes. My only qualification is to replace "an empirical approach"
with the plural "empirical approaches". (019)
I'll summarize all this discussion with a quotation by C. S. Peirce:
> It is easy to speak with precision upon a general theme. Only, one must
> commonly surrender all ambition to be certain. It is equally easy to be
> certain. One has only to be sufficiently vague. It is not so difficult
> to be pretty precise and fairly certain at once about a very narrow subject. (020)
John (021)
_________________________________________________________________
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 (022)
|