From: MR
To: Editor
Date: 3 Apr 2005
Do you have any papers on the Semantic Web and how it will
suck horribly if it uses XML?
I'm working on my Master and one of the courses I'm taking is
a seminar on XML/Semantic Web. I have to make a 30-minute presentation on some
topic of the Semantic Web. Most of the students are going to present
XML-lovefest fluff pieces and, since I took the course to try and gauge the
amount of nonsense in my particular piece of academia, I'd rather have
something which challenges some of the core assumptions of the Semantic Web
(namely that XML/RDF are sufficient and correct ways to implement it).
Unfortunately, Google searching for Semantic Web and Debunk
or Criticism is highly lacking. The "best" I've gotten so far is this
Shirky piece, The Semantic Web, Syllogism, and Worldview, but itself has
been somewhat debunked in Themes and Metaphors in the Semantic Web
Discussion. One article I found was A Response to Clay Shirky. In
it, Paul Ford explains how a "Relational Database" wouldn't be able
to do this. It's obvious that he's thinking of SQL DBMS, so my thoughts are to
debunk his ContraShirky post and illustrate how a TRDBMS would be able to do
this.
However, if you have any pieces which would prove to be
either 1) more authoritative or 2) more correct or 3) more interesting or 4)
something else, I'd gladly pick a PRACTICAL
DATABASE FOUNDATIONS paper up. The Un-Muddling Modeling
paper seems to be somewhat relevant, but not having seen the contents I'm not
sure yet.
From: Fabian Pascal
To: MR
No, because no longer being a student, I have some control
over the crap that I want to deal with and I always go over the quota. When I
was a student, I had my share of that, but it was nowhere near the amount of
crap in academia these days. The corruption by industry and ensuing dumbing
down has practically destroyed it, and it's irreversible (see More on the
Chasing of Mayflies, forthcoming).
My suggestion is to come up with what semantics means from
some good logic sources e.g. MEANING AND ARGUMENT: ELEMENTS OF LOGIC by
Robert Olson (oldie but goodie and a rare case of accessible logic book), then
figure out what Semantic Web proponents mean by it, and compare both to what
XML provides. You should see the nonsense of it. There are a couple of XML Links on this site that
are also useful.
You're not going to find much of value if you limit yourself
to current industry material, even if it's by academics, because today those
are no longer really academics, but panderers to the industry on the one hand,
or students who only want to be trained for a job, not to be educated, on the
other.
Think of where the meaning of data in databases
resides. Is XML equivalent? Think of what a data model is and what its
purpose is. How does XML compare?
Incidentally, I am writing some papers analyzing Codd's
original work, and here's a paragraph that is pertinent:
Note that column labels "partially convey
significance" does not imply that there is meaning in the names. The names
serve as an incomplete (hence "partial") carrier of the distinction
between columns: they signify a difference, but do not explain to a system what
it means (which they would have to if they were to convey meaning). This is
another way of saying that meaning should not be “hidden” from the DBMS in
names.
From: MR
Unfortunately the profs require that it be some sort of
"published" paper/work so I can't really invent my own even if it's
the right thing to do.
Thanks for your help; I see that Logic book is $5 on various
out-of-print bookstores--I'll pick up a copy. The semantic web uses some sort
of layer on top of XML called "RDF". In short, it looks an awful lot
like the "Associative Model of Data"--everything is represented as a
triple "something isa somethingelse" etc. The whole idea is that they
would only use the Semantic Web as a glorified "search" tool. Namely,
you would only want to know what bits of text relate to other bits of text. It
seems somewhat shortsighted to me, but it's apparently the new fad in the XML
camp(s) and is unfortunately spreading.
From: Fabian Pascal
Sure, but my point was not only material published currently.
Go back to material published years ago, when authors still knew something
beyond the latest industry buzzwords that mean absolutely nothing.
As a student you should read such books not just to satisfy
the profs, but to learn something that those profs may not teach you. The book
I suggested is well beyond what you need for this paper, but it’s material any
practitioner should know.
You're on the right track. If you know what a data model is
(e.g. relational model), it has three components. Those three components together
convey meaning. Find out what the XML equivalents, if any, are; and what
Semantic Web proponents have in their mind as to what meaning means. That'll tell
you what they'll end up with, and if it can do more or less than glorified
searches.
Ed. Note: Somebody
did a MSN search on ‘meaning database’ (possibly MR). It’s enlightening, given
the subject of this exchange to inspect the results:
http://search.msn.com/results.aspx?q=meaning%20of%20database&FORM=QBRE
Posted 6/3/05
