Dear Matthew, (01)
I more or less agree with your point: (02)
> MW: I still disagree. Traditional systems may
> or may not have a semantic interpretation. That
> depends on the design approach that was taken. (03)
An SQL database, for example, has a very clear
model-theoretic semantics: each relation is a
collection of tuples that give an extensional
definition of the relation. The WHERE clause of
a relational DB is a first-order expression whose
truth value is evaluated in terms of the relations.
The method of doing the evaluation is consistent
with Tarski's model theory (given the assumption
of negation as failure). (04)
However, SQL does have nasty properties, such as
null values, which were an afterthought that many
of the original developers strongly regretted. (05)
There is also the possibility (very real in many
cases) that the original DB administrator/analyst
may not have made wise choices in the translation
of the application data into the tuples of the RDB. (06)
Unfortunately, the same weaknesses are also true
of the way people are using RDF and OWL. In many
cases, the RDF & OWL users have less training in
design principles, and their designs are far less
reliable than the good RDB designs of the 1970s. (07)
I have been a strong critic of SQL and UML for many
years, but after looking at some of the more recent
designs by people who claim to be doing "ontology",
I have much more sympathy with the old-time DB
administrators. Many of them were highly trained
professionals with long years of experience. They might
not have had a strong background in logic, linguistics,
and philosophy, but the good ones were very good. (08)
John Sowa (09)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/uos-convene/
To Post: mailto:uos-convene@xxxxxxxxxxxxxxxx
Community Portal: http://ontolog.cim3.net/
Shared Files: http://ontolog.cim3.net/file/work/UpperOntologySummit/uos-convene/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?UpperOntologySummit (010)
|