Martin, (01)
Thanks for the pointers to your papers. I agree with a lot of what
you say, but my major concern is that there is much, much more to say
that goes far beyond what the Semantic Web is currently doing. A lot
of useful work has been done within the SemWeb paradigm, but my major
complaint is that it's too provincial. The foundation was established
without considering the immense amount of mission-critical technology
that was already available and very widely used. (02)
Relational databases are the most obvious technology that was omitted.
The world economy runs on RDBs, and nearly every major commercial web
site is integrated with an RDB. For smaller sites, the standard is
LAMP: Linux, Apache, MySQL, and Perl, Python, or PHP. (03)
When RDF(S) was being designed, the obvious approach was to support
arbitrary n-tuples. That would have made it trivial to download or
upload any or all of an SQL database (or the response from an SQL
query) to or from RDF n-tuples. Some people who had a bright idea
for implementing "triple stores" thought that they could implement
triples faster than n-tuples. But that is not only false, it
would be hopelessly nearsighted as a design decision, even if it
had a grain of truth. (04)
Another weakness is that the designers ignored the major lesson
of DB design from the earliest, pre-relational days: a DB without
indexing is hopelessly inefficient. Embedding data in web pages
is useful for many purposes, but high-speed processing of large
volumes of data requires downloading and indexing. (Note Google.) (05)
Another limitation is caused by ignoring existing standards.
There is an ISO standard for Prolog, but people are still working
on an incredibly underpowered version called RuleML. Many major
sites routinely download RDF and OWL into Prolog in order to get
any kind of reasonable performance. At our company (VivoMind),
we download and translate RDF and OWL into Prolog faster than
most native processors can do just the download. Then the
Prolog version runs circles around the native systems. (06)
Prolog is the major language used for immensely large systems
that do complex inferences. As just one example, Experian
(one of the three major credit bureaus that check everybody's
credit worthiness) uses Prolog for all their complex checks.
In fact, they use it so much that they bought the Prologia
company, which was founded by Alain Colmerauer, the person
who first designed and implemented Prolog. (07)
There is much more to say, but it's essential for the SemWebbers
to recognize that there is an enormous amount of very important
theory and technology that must be integrated with the Sem Web
before it can reach the high hopes that people had for it. And
integration does *not* mean a one-sided mapping of everything
outside the Sem Web into XML-based notations. (08)
And by the way, I'm pleased that you liked my paper "Fads and
Fallacies about Logic." Jim Hendler was the editor of the
journal in which it was published, and Jim liked the paper
-- despite the fact that he and I have had many arguments
over the years about many things (although we occasionally
agree about a few). (09)
John (010)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Config: 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 Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (011)
|