On May 22, 2007, at 6:11 AM, Marijke Keet wrote:
<snip>
The history of computing is the history of "design patterns" at
one level that eventually get built into "higher level
languages" at the next level of abstraction up.
I think I have a less optimistic view of progress in computer
science. For example, many of the paradigmatic GoF design
patterns are there to make up for deficiencies in the OO
languages that *succeeded* more expressive and abstract
functional languages.
Amen to that. And we are living through an exactly similar
transition in representational languages, where DLs are re-
inventing axioms of classical logic.
it's not quite re-inventing, although it may be that "new features
added to a language" are sometimes being sold as if they were
novel. As for the n-aries in DLs (which are indeed trivial in CL),
that is possible in DLs in theory for over 10 years and in software
with iCOM for >7 years -- and do the automated reasoning over it,
unlike with several other logics.
I like more expressivity as well, but then, I'm not implementing
systems where I'd have to wait 'long' for query answers or see my
computer hang upon classifying 1 instance in an 50-concept small
ontology (with the latest pellet for owl 1.1). I did try to load in
Protégé and SWOOP the FMA-lite, which is a 43MB OWL file. It failed.
Actually, I have to defend the OWL tools here, there are known
problems with some of the relations in FMA and FMA-lite. These cause
simple forward chaining reasoning to explode and I'm guessing the
same thing happens with tableau algorithms.. the FMA developers are
aware, these are being incrementally fixed in the main FMA and will
hopefully all be fixed in the next release of FMA_lite in a month or so
The problems are of the sort
ALL rectum part_of male pelvis
ALL rectum part_of female pelvis
(but more widespread and deeper down the partonomy)
I'm guessing this is at least contributing to your problem
Thanks to Stuart Aitken for noticing this problem
Reasoning over sections of the FMA that take into account only some
constructors is possible [1], which brings us back to your earlier
comment that "people have argued against more expressive languages,
in fact have argued with great force and vehemence,": if we have to
chop up large ontologies anyway in order to be able to reason over
them, we might as well do that in a structured manner with some
simpler languages and (semi-)automated conversions for "dumbing
down" a large and/or rich ontology to some slimmed version that is
computationally tractable; that is, taking best of 'both worlds'
with expressivity where desired/needed and performance where needed/
desired.
[1] Zhang S, Bodenreider O, Golbreich C. Experience in reasoning
with the Foundational Model of Anatomy in OWL-DL. In:Pacific
Symposium on Biocomputing 2006, Altman RB, Dunker AK, Hunter L,
Murray TA, Klein TE, (Eds.). World Scientific, 2006, 200-211.
http://helix-web.stanford.edu/psb06/zhang_s.pdf
regards,
marijke
C. Maria Keet
KRDB Research Centre
Faculty of Computer Science
Free University of Bozen-Bolzano
Piazza Domenicani 3
39100 Bozen-Bolzano
Italy
tel: +39 04710 16128
fax: +39 04710 16009
email: [EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>
web: http://www.inf.unibz.it/krdb/
home: http://www.meteck.org <http://www.meteck.org/>