From: PK
To: Editor
Date: 10/11/2003
In my spare time I'm researching a mechanism to translate a
semantic temporal requirements specification language (R.K. Stamper's LEGOL) to
the relational model. I came across Mr. Codd's article Extending the
Database Relational Model to Capture More Meaning and found RM/T a very
useful approach for my work. Since Mr. Codd did not provide example relvars for
what he defined in the paragraphs 11 "generalization",12 "cover aggregation" and
13 "event precedence", I'm not quite sure if I really understood RM/T in depth.
In Mr. Date's books AN INTRODUCTION TO DATABASE SYSTEMS
7th Ed. and TEMPORAL DATA AND THE RELATIONAL MODEL only some
polite endnotes on RM/T are spent. So I tried to find sources on RM/T in my
local university library, which had nothing, and the Internet. Maybe due to the
fact that "Google" ignores slashes, I did not find many relevant
articles. At least I found out, that Mr. Date has written a section in the book
RELATIONAL DATABASE WRITINGS 1991-1994 that I ordered via Amazon.com,
but they already postponed the delivery once and I am not sure I will get the
book at all. Most other sources I found, such as lecturer notes, do merely
quote Mr. Codd's article. One German researcher seems to have
"modified" RM/T, but I cannot get this publication. Someone on the
Internet claimed, Mr. Codd had merged RM/T into his RM/V2, but in his book THE
RELATIONAL MODEL FOR DATABASE MANAGEMENT: VERSION 2 I have found nothing
whatsoever about surrogates, entity relations etc. that make up RM/T.
So my question to you, if you are willing to spend the time
for a reply, is: What happened to RM/T? Was it too good or too bad to receive
serious follow-up? Could you give me any sources for further study, or is there
a good reason to not consider RM/T in any case?
Chris Date Responds: As far as I know, neither Codd
nor anybody else ever did much to follow up on the original (1979) RM/T paper.
You're right in thinking nothing from RM/T found its way into RM/V2. As for the
coverage in my own book RELATIONAL DATABASE
WRITINGS 1991-1994, I can't in honesty recommend it to someone like
yourself who has already studied the subject--it's just a short tutorial, and
it says little or nothing about the specific topics you mention
(generalization, cover aggregation, and event precedence). And, sadly, Codd himself died earlier
this
year, so we can't ask him about such matters any more.
That said, I want to add that I'm not at all sure it's a good
idea for you to try to base your research too exclusively on RM/T. The informal
RM/T classification of entities into kernels, characteristics, and associations
is helpful, at least from an intuitive point of view. But, frankly, I don't
think the RM/T notions of generalization, cover aggregation, and event
precedence were ever very clear, and--as I've already indicated--they haven't
been elaborated since, so far as I know. The same is true of the RM/T operators
as well. What's more, the RM/T paper was the first in which Codd seriously
advocated the use of NULLs and three-valued logic, and--along with many
others--I've been opposed to those ideas for many years! I seriously hope you're not making
use of
NULLs in your work ...
Regarding "generalization" in particular, I think
we have a much better understanding now than we did back in 1979 of what this
notion really involves. Hugh Darwen and I have presented an extensive description of how we think
"generalization"can be realized in a relational context in our book FOUNDATION FOR FUTURE
DATABASE SYSTEMS: THE THIRD MANIFESTO (2nd ed.) Our presentation
emphasizes "generalization" as it applies to types (= domains) rather
than relvars, but it does show how to bring relvars too into the
"generalization" fold (as it were).
Note: I've heard that people have had difficulty obtaining
the MANIFESTO book. Let me therefore mention that a tutorial discussion of the same material
can be found in AN INTRODUCTION TO
DATABASE SYSTEMS,
(8th ed).
Regarding "events" and "event
precedence," I think we have a better understanding of this topic too now
than we did in 1979. I'm glad you seem
to be familiar with the book TEMPORAL DATA AND THE
RELATIONAL MODEL, because that book contains our current best thinking
on this topic. I would hope that the ideas it contains would be applicable to
your research. Indeed, if you encounter some specific problems in your research where our ideas
don't seem to work, we'd be very interested to hear about them.
I wish you success in your continuing efforts in this field.
Posted
11/14/03
[ABOUT]
[QUOTES]
[LINKS]
