Re: [tlug] Effect of AI and data processing rift on the Semantic Web

[Curt Sampson <cjs@example.com>]

Ok, time for a rant.

On 2008-09-02 09:59 +0900 (Tue), Edward Middleton wrote:

> This[1] is an interesting post about the historical division between the 
> data processing and AI camps, and discusses how it has negatively effected 
> the development of the semantic web.

Actually, the post entirely misses the real division. It's nothing to do
with what machines you're running on, or whether someone uses COBOL or
not. It's a very simple and long-standing division that persists to this
day.

If you want to see it, go around to all of the DBAs you know, and ask
them what a relational database management system is. I'll bet you a
beer, no two, that you get a lot of "an engine for storing and searching
data"-type responses, and not one of them comes back with, "a logical
inference engine."

As Sowa pointed out in the post you referenced:

] Now the *world* economy runs on RDBs.
]
] Almost every major web site includes a large RDB, and the medium
] sized web sites are based on LAMP:  Linux, Apache, MySQL, and
] Perl, Python, or PHP.

This mention of MySQL is telling, since it's pretty much the most
anti-relational RDBMS out there: an inference engine isn't much good if
the facts it's working from are junk, and the MySQL developers (and many
of its users) have persisted for years in saying that MySQL accepting
"February 31st", or even "Februry 0th", as a date is a fine thing.
These people are actively opposed to data having machine-interpretable
meaning!

And then there's all those object-oriented folks who, not content with
having removed half of the good stuff an RDBMS should give us, are
busy writing "object-relational mapping layers" in an attemp (and an
apparently successful one, I might add) to remove the other half, and
bring us back to the bad old days of network databases, the problems of
which relational systems were desgined to fix in the first place.

It doesn't help that the XML zealots have managed to drag our data
modeling a good way back in that direction, either.

And of course SQL was a mistake in the first place, but System/R, for
whatever reasons (ask Hugh Darwen), did better in the commerical market
than Business System 12, and now we're in a state where even if you
do understand relational theory, and have a good RDBMS to work with,
translating your nice relational query into SQL varies in difficulty
from painful to impossible.

Basically, most of the data-processing world considers an RDBMS to be
merely a place to dump a big mess of data, and that's it. People like
these are not friends of logic or of semantic anything.

cjs
-- 
Curt Sampson       <cjs@example.com>        +81 90 7737 2974   
Mobile sites and software consulting: http://www.starling-software.com