Leo, (01)
I don't know how you or Bill define "reasoning system",
but if you define a typical expert system, as the term
was used from the 1980s to today, as a reasoning system,
then the same term is applicable to a relational DBMS. (02)
> Oh, come on, Bill. Since when are relational databases
> reasoning systems? (03)
The collection of tuples that constitutes the data
can be considered a collection of ground-level clauses. (04)
The SQL WHERE clause in a query or constraint allows
you write an arbitrary first-order expression that
is evaluated against the collection of tuples. (05)
The "views" that define "virtual relations" in terms of
stored relations or other virtual relations are comparable
to Prolog rules that define new predicates. They don't
give you the full power of Prolog, but they do give you
a backward-chaining Prolog-like subset (comparable to
Datalog). (06)
And the "triggers" that are supported by most of the
commercial DBMSs give you a forward-chaining system that
is of the same nature as an OPS5 (or CLIPS) system.
(By "same nature", I mean similar in the way rules are
triggered, but it is not as general or as optimized as CLIPS.) (07)
And if the views and triggers don't give you all the
power necessary for an application, those systems do
support procedural attachments. (08)
I will certainly admit that I would prefer a more
systematic and less quirky system than what is provided
by a typical RDBMS. But could anyone claim that RDF
and OWL are less quirky than SQL? (09)
John (010)
_________________________________________________________________
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 (011)
|