Re: SEMIS Bulletin
Jim,
Re debate at ICCS: I am indeed sorry that we will
miss one another there.
Re contacting Pat: This thread continued on for
a week or so, and I cc'd Pat after the first few
interchanges. He has always felt that the semantic
web support for logic-based work was good, and I
agree with that -- with many, many caveats. I'm
putting Pat on the cc list for this note so that
he can put his own inimitable emphasis on any of
these points, if he is so inclined.
Re RDF: Something like that is needed, and I recommend
using it for the same reason that Harlan Mills gave
for OS/360 about 35 years ago:
Any decent computer scientist can think of a million
ways to improve OS/360, but the best way to treat it
is to compare it to a cow. If you put hay and water
in one end, you get fertilizer out the other end and
milk from the middle. Just keep your ends straight,
and it works.
That is a cute analogy, but it fails to answer the
computer scientists. There were some very bad design
decisions, which I and many others criticized at the
time they were being made. The current version of
IBM's Z/OS, which has been rewritten many times over
since it grew out of OS/360, resembles the Pentium
architecture, which still has the buried skeletons
of the Intel 4004 (their original 4-bit CPU). But
RDF & OWL are already kludges, and they haven't even
progressed to the 8008 stage.
Re 30-year-old technology: As I have said many times
before, there is nothing wrong with using 30-year-old
or even 2000-year-old technology. My reason for
pointing out the age (or maturity) of the technology
is that all the major design decisions had been
thoroughly explored, and there was no excuse for
designing a kludge that ignored SQL, UML, & EXPRESS.
For example, instead of triples, the W3C could have
adopted tuples, and that decision would have facilitated
the integration of RDF with the immense amount of technology
built for SQL, Express, and UML.
JH> examples - more people have Foaf files (estimated 2-5M users)
> than had web pages 6 years in Adobe is putting RDF (XMP) in every
> PDF and photoshop document
> a recent sem web search engine has found RDF or OWL on 250k
> web pages (total of over 20M triples) some "intranet" applications
> have been reported in govt applications with over 1B triples in
> various triple stores So in fact, we are doing extremely well -
> compared to the pre-Mosaic web, we are way ahead and growing well
As I said, those statistics are insignificant compared to the
existing SQL, UML, and EXPRESS technology. 20M triples is trivial
for any modern database, even for a desktop system. What I said
is that the greatest strength of the original WWW is the way it
built on existing technology. And the greatest weakness of the
semantic web is that it ignored existing technology. (Yes, many
people are combining RDF & OWL with existing technology, but
they have to start from scratch without any guidelines.)
My major complaint is that instead of paving the way for better
systems in the future, RDF and OWL are just two more legacy
systems that we will have to work around.
John